{"schemaVersion":3,"runs":[{"id":"adaptive-planning-original-2026-09","categoryId":"workday-adaptive-planning","title":"Original technical evidence 1","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"deepseek/deepseek-v4.1-flash","modelName":"DeepSeek: DeepSeek V4.1 Flash","provider":"InferenceNet","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-I01"],"heldQuestionIds":[],"completedCount":19,"assessedCount":19,"attemptCount":21,"statusCounts":{"completed":19,"truncated":1,"api-error":0,"transport-error":0,"protocol-error":1,"empty":0,"interrupted":0},"qualityScore":77.89473684210526,"medianLatencyMs":227853,"knownCostUsd":0.022724465,"unknownCostAttemptCount":0,"totalCostUsd":0.022724465,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"z-ai/glm-5.3-flash","modelName":"Z.ai: GLM 5.3 Flash","provider":"Z.AI","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":65,"medianLatencyMs":51685,"knownCostUsd":0.01594333,"unknownCostAttemptCount":0,"totalCostUsd":0.01594333,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"tencent/hy4-preview","modelName":"Tencent: Hy4 preview","provider":"Tencent","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F08"],"heldQuestionIds":[],"completedCount":1,"assessedCount":1,"attemptCount":21,"statusCounts":{"completed":1,"truncated":17,"api-error":0,"transport-error":2,"protocol-error":0,"empty":0,"interrupted":1},"qualityScore":30,"medianLatencyMs":178167,"knownCostUsd":0.366827497,"unknownCostAttemptCount":1,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"openai/gpt-5.6-luna","modelName":"OpenAI: GPT-5.6 Luna","provider":"Azure","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-I01"],"heldQuestionIds":[],"completedCount":20,"assessedCount":20,"attemptCount":21,"statusCounts":{"completed":20,"truncated":1,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":73.75,"medianLatencyMs":15880.5,"knownCostUsd":0.0784328,"unknownCostAttemptCount":0,"totalCostUsd":0.0784328,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"deepseek/deepseek-v4-flash-0731","modelName":"DeepSeek: DeepSeek V4 Flash 0731","provider":"Makora","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":66.42857142857143,"medianLatencyMs":10877,"knownCostUsd":0.00958899,"unknownCostAttemptCount":0,"totalCostUsd":0.00958899,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"xiaomi/mimo-v2.5","modelName":"Xiaomi: MiMo-V2.5","provider":"Requested only (not observed): DeepInfra","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":[],"heldQuestionIds":[],"completedCount":0,"assessedCount":0,"attemptCount":21,"statusCounts":{"completed":0,"truncated":0,"api-error":20,"transport-error":1,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":null,"medianLatencyMs":null,"knownCostUsd":0.000507661,"unknownCostAttemptCount":20,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"deepseek/deepseek-v4-flash","modelName":"DeepSeek: DeepSeek V4 Flash 0423","provider":"Venice","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":65.71428571428571,"medianLatencyMs":29676,"knownCostUsd":0.004762954,"unknownCostAttemptCount":0,"totalCostUsd":0.004762954,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"tencent/hy3","modelName":"Tencent: Hy3","provider":"Tencent","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F02","AP-F03","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C06","AP-C07","AP-C08","AP-C09","AP-I01"],"heldQuestionIds":[],"completedCount":13,"assessedCount":13,"attemptCount":21,"statusCounts":{"completed":13,"truncated":8,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":71.92307692307692,"medianLatencyMs":37332,"knownCostUsd":0.0440765325,"unknownCostAttemptCount":0,"totalCostUsd":0.0440765325,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"z-ai/glm-5.3","modelName":"Z.ai: GLM 5.3","provider":"Z.AI","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":74.28571428571429,"medianLatencyMs":18863,"knownCostUsd":0.12607444,"unknownCostAttemptCount":0,"totalCostUsd":0.12607444,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"google/gemini-3.8-flash","modelName":"Google: Gemini 3.8 Flash","provider":"Google","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F03","AP-F04","AP-F05","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":19,"assessedCount":19,"attemptCount":21,"statusCounts":{"completed":19,"truncated":2,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":78.42105263157895,"medianLatencyMs":14897,"knownCostUsd":0.2428035,"unknownCostAttemptCount":0,"totalCostUsd":0.2428035,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"openai/gpt-6-luna","modelName":"OpenAI: GPT-6 Luna","provider":"Azure","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":74.04761904761905,"medianLatencyMs":13446,"knownCostUsd":0.0261795,"unknownCostAttemptCount":0,"totalCostUsd":0.0261795,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"upstage/solar-pro4","modelName":"Upstage: Solar Pro 4","provider":"Upstage","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":20,"assessedCount":20,"attemptCount":21,"statusCounts":{"completed":20,"truncated":0,"api-error":0,"transport-error":1,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":66.75,"medianLatencyMs":27209,"knownCostUsd":0.01870533,"unknownCostAttemptCount":0,"totalCostUsd":0.01870533,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"z-ai/glm-5.2","modelName":"Z.ai: GLM 5.2","provider":"Z.AI","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F02","AP-F04","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-I01"],"heldQuestionIds":[],"completedCount":16,"assessedCount":16,"attemptCount":21,"statusCounts":{"completed":16,"truncated":5,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":74.6875,"medianLatencyMs":18756.5,"knownCostUsd":0.34630252,"unknownCostAttemptCount":0,"totalCostUsd":0.34630252,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"openai/gpt-5.6-sol","modelName":"OpenAI: GPT-5.6 Sol","provider":"Azure","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":83.57142857142857,"medianLatencyMs":15202,"knownCostUsd":0.56622,"unknownCostAttemptCount":0,"totalCostUsd":0.56622,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"anthropic/claude-sonnet-5","modelName":"Anthropic: Claude Sonnet 5","provider":"Amazon Bedrock","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":66.9047619047619,"medianLatencyMs":11922,"knownCostUsd":0.247086,"unknownCostAttemptCount":0,"totalCostUsd":0.247086,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"minimax/minimax-m3","modelName":"MiniMax: MiniMax M3","provider":"Together","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":20,"assessedCount":20,"attemptCount":21,"statusCounts":{"completed":20,"truncated":1,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":62.25,"medianLatencyMs":10369,"knownCostUsd":0.04424238,"unknownCostAttemptCount":0,"totalCostUsd":0.04424238,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"moonshotai/kimi-k3","modelName":"MoonshotAI: Kimi K3","provider":"Moonshot AI","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F06","AP-F07","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":19,"assessedCount":19,"attemptCount":21,"statusCounts":{"completed":19,"truncated":0,"api-error":0,"transport-error":2,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":86.3157894736842,"medianLatencyMs":61583,"knownCostUsd":0.663354,"unknownCostAttemptCount":0,"totalCostUsd":0.663354,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"inclusionai/ling-3.0-flash-fin:free","modelName":"inclusionAI: Ling 3.0 Flash Fin (free)","provider":"Novita","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08"],"heldQuestionIds":[],"completedCount":9,"assessedCount":9,"attemptCount":21,"statusCounts":{"completed":9,"truncated":0,"api-error":12,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":71.66666666666667,"medianLatencyMs":4323,"knownCostUsd":0,"unknownCostAttemptCount":12,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"openai/gpt-6-astra","modelName":"OpenAI: GPT-6 Astra","provider":"Azure","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":20,"assessedCount":20,"attemptCount":21,"statusCounts":{"completed":20,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":1,"empty":0,"interrupted":0},"qualityScore":85,"medianLatencyMs":20571.5,"knownCostUsd":1.34278,"unknownCostAttemptCount":1,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]},{"modelId":"deepseek/deepseek-v4-pro-0813","modelName":"DeepSeek: DeepSeek V4 Pro 0813","provider":"Ionstream","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":18,"assessedCount":18,"attemptCount":21,"statusCounts":{"completed":18,"truncated":2,"api-error":0,"transport-error":0,"protocol-error":0,"empty":1,"interrupted":0},"qualityScore":71.66666666666667,"medianLatencyMs":52842,"knownCostUsd":0.190704832,"unknownCostAttemptCount":0,"totalCostUsd":0.190704832,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-01-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 1","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"openai/gpt-6-astra","modelName":"OpenAI: GPT-6 Astra","provider":"Azure","questionIds":["AP-F07"],"assessedQuestionIds":["AP-F07"],"heldQuestionIds":[],"completedCount":1,"assessedCount":1,"attemptCount":1,"statusCounts":{"completed":1,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":40,"medianLatencyMs":28258,"knownCostUsd":0.06777,"unknownCostAttemptCount":0,"totalCostUsd":0.06777,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-02-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 2","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"moonshotai/kimi-k3","modelName":"MoonshotAI: Kimi K3","provider":"Moonshot AI","questionIds":["AP-F05","AP-F08"],"assessedQuestionIds":["AP-F05","AP-F08"],"heldQuestionIds":[],"completedCount":2,"assessedCount":2,"attemptCount":2,"statusCounts":{"completed":2,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":95,"medianLatencyMs":57144.5,"knownCostUsd":0.05904,"unknownCostAttemptCount":0,"totalCostUsd":0.05904,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-03-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 3","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"z-ai/glm-5.2","modelName":"Z.ai: GLM 5.2","provider":"Z.AI","questionIds":["AP-F01","AP-F03","AP-F05","AP-C04","AP-C10"],"assessedQuestionIds":["AP-F01","AP-F03","AP-F05","AP-C04","AP-C10"],"heldQuestionIds":[],"completedCount":5,"assessedCount":5,"attemptCount":5,"statusCounts":{"completed":5,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":68,"medianLatencyMs":106766,"knownCostUsd":0.1719002,"unknownCostAttemptCount":0,"totalCostUsd":0.1719002,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-04-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 4","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"google/gemini-3.8-flash","modelName":"Google: Gemini 3.8 Flash","provider":"Google","questionIds":["AP-F02","AP-F06"],"assessedQuestionIds":["AP-F02","AP-F06"],"heldQuestionIds":[],"completedCount":2,"assessedCount":2,"attemptCount":2,"statusCounts":{"completed":2,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":72.5,"medianLatencyMs":36441.5,"knownCostUsd":0.035688,"unknownCostAttemptCount":0,"totalCostUsd":0.035688,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-05-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 5","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"tencent/hy4-preview","modelName":"Tencent: Hy4 preview","provider":"Tencent","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C06"],"heldQuestionIds":[],"completedCount":14,"assessedCount":14,"attemptCount":16,"statusCounts":{"completed":14,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":1,"empty":0,"interrupted":1},"qualityScore":74.64285714285714,"medianLatencyMs":439119,"knownCostUsd":0.543832762,"unknownCostAttemptCount":1,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-06-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 6","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"deepseek/deepseek-v4-pro-0813","modelName":"DeepSeek: DeepSeek V4 Pro 0813","provider":"Ionstream","questionIds":["AP-F01","AP-C04","AP-C05"],"assessedQuestionIds":["AP-F01","AP-C04","AP-C05"],"heldQuestionIds":[],"completedCount":3,"assessedCount":3,"attemptCount":3,"statusCounts":{"completed":3,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":86.66666666666667,"medianLatencyMs":68198,"knownCostUsd":0.031684112,"unknownCostAttemptCount":0,"totalCostUsd":0.031684112,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-07-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 7","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"openai/gpt-5.6-luna","modelName":"OpenAI: GPT-5.6 Luna","provider":"Azure","questionIds":["AP-C10"],"assessedQuestionIds":["AP-C10"],"heldQuestionIds":[],"completedCount":1,"assessedCount":1,"attemptCount":1,"statusCounts":{"completed":1,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":60,"medianLatencyMs":37547,"knownCostUsd":0.0064874,"unknownCostAttemptCount":0,"totalCostUsd":0.0064874,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-08-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 8","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"minimax/minimax-m3","modelName":"MiniMax: MiniMax M3","provider":"Together","questionIds":["AP-F06"],"assessedQuestionIds":["AP-F06"],"heldQuestionIds":[],"completedCount":1,"assessedCount":1,"attemptCount":1,"statusCounts":{"completed":1,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":75,"medianLatencyMs":13142,"knownCostUsd":0.00213438,"unknownCostAttemptCount":0,"totalCostUsd":0.00213438,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-09-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 9","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"tencent/hy3","modelName":"Tencent: Hy3","provider":"Tencent","questionIds":["AP-F01","AP-F04","AP-F05","AP-F06","AP-C03","AP-C04","AP-C05","AP-C10"],"assessedQuestionIds":["AP-F01","AP-F04","AP-F05","AP-F06","AP-C03","AP-C04","AP-C05","AP-C10"],"heldQuestionIds":[],"completedCount":8,"assessedCount":8,"attemptCount":8,"statusCounts":{"completed":8,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":72.5,"medianLatencyMs":94749,"knownCostUsd":0.03192189,"unknownCostAttemptCount":0,"totalCostUsd":0.03192189,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-10-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 10","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"upstage/solar-pro4","modelName":"Upstage: Solar Pro 4","provider":"Upstage","questionIds":["AP-C06"],"assessedQuestionIds":["AP-C06"],"heldQuestionIds":[],"completedCount":1,"assessedCount":1,"attemptCount":1,"statusCounts":{"completed":1,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":55,"medianLatencyMs":28261,"knownCostUsd":0.00092574,"unknownCostAttemptCount":0,"totalCostUsd":0.00092574,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-11-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 11","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"deepseek/deepseek-v4.1-flash","modelName":"DeepSeek: DeepSeek V4.1 Flash","provider":"Requested only (not observed): InferenceNet","questionIds":["AP-C09","AP-C10"],"assessedQuestionIds":[],"heldQuestionIds":[],"completedCount":0,"assessedCount":0,"attemptCount":2,"statusCounts":{"completed":0,"truncated":0,"api-error":2,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":null,"medianLatencyMs":null,"knownCostUsd":0,"unknownCostAttemptCount":2,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-12-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 12","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"xiaomi/mimo-v2.5","modelName":"Xiaomi: MiMo-V2.5","provider":"DeepInfra","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F04","AP-F05"],"heldQuestionIds":[],"completedCount":3,"assessedCount":3,"attemptCount":8,"statusCounts":{"completed":3,"truncated":0,"api-error":4,"transport-error":1,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":60,"medianLatencyMs":424327,"knownCostUsd":0.004671758,"unknownCostAttemptCount":4,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-recovery-13-2026-09","categoryId":"workday-adaptive-planning","title":"Recovery technical evidence 13","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"inclusionai/ling-3.0-flash-fin:free","modelName":"inclusionAI: Ling 3.0 Flash Fin (free)","provider":"Novita","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":12,"assessedCount":12,"attemptCount":12,"statusCounts":{"completed":12,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":62.5,"medianLatencyMs":4678,"knownCostUsd":0,"unknownCostAttemptCount":0,"totalCostUsd":0,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-supplement-2026-09","categoryId":"workday-adaptive-planning","title":"Supplement technical evidence 1","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"stealth/space-bunny-alpha","modelName":"Space Bunny Alpha","provider":"Stealth","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":21,"assessedCount":21,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":66.42857142857143,"medianLatencyMs":34114,"knownCostUsd":0,"unknownCostAttemptCount":0,"totalCostUsd":0,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-completion-deepseek-v41-flash-2026-09","categoryId":"workday-adaptive-planning","title":"Completion technical evidence 1","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"deepseek/deepseek-v4.1-flash","modelName":"DeepSeek: DeepSeek V4.1 Flash","provider":"InferenceNet","questionIds":["AP-C09","AP-C10"],"assessedQuestionIds":["AP-C09","AP-C10"],"heldQuestionIds":[],"completedCount":2,"assessedCount":2,"attemptCount":2,"statusCounts":{"completed":2,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":42.5,"medianLatencyMs":73662.5,"knownCostUsd":0.00269675,"unknownCostAttemptCount":0,"totalCostUsd":0.00269675,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-completion-hy4-2026-09","categoryId":"workday-adaptive-planning","title":"Completion technical evidence 2","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"tencent/hy4-preview","modelName":"Tencent: Hy4 preview","provider":"Tencent","questionIds":["AP-C05","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-C05","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":6,"assessedCount":6,"attemptCount":6,"statusCounts":{"completed":6,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":75,"medianLatencyMs":354903,"knownCostUsd":0.19231383,"unknownCostAttemptCount":0,"totalCostUsd":0.19231383,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-completion-mimo-v25-2026-09","categoryId":"workday-adaptive-planning","title":"Completion technical evidence 3","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"xiaomi/mimo-v2.5","modelName":"Xiaomi: MiMo-V2.5","provider":"DeepInfra","questionIds":["AP-F02","AP-F03","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F02"],"heldQuestionIds":[],"completedCount":1,"assessedCount":1,"attemptCount":4,"statusCounts":{"completed":1,"truncated":0,"api-error":3,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":60,"medianLatencyMs":39596,"knownCostUsd":0.000197638,"unknownCostAttemptCount":3,"totalCostUsd":null,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]},{"id":"adaptive-planning-completion-mimo-novita-2026-09","categoryId":"workday-adaptive-planning","title":"Completion technical evidence 4","summary":"Technical evidence retained for coverage and audit; batch quality uses first-completed answers.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","productVersion":"Unknown release; no tenant validation","track":"model-only","environment":"The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","questionCount":21,"reviewStatus":"approved","qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"This technical source is combined into the batch; its model rows are not separate comparisons.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"models":[{"modelId":"xiaomi/mimo-v2.5","modelName":"Xiaomi: MiMo-V2.5","provider":"Novita","questionIds":["AP-F03","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F03","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"completedCount":17,"assessedCount":17,"attemptCount":17,"statusCounts":{"completed":17,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"qualityScore":55,"medianLatencyMs":23105,"knownCostUsd":0.008986824,"unknownCostAttemptCount":0,"totalCostUsd":0.008986824,"comparisonStatus":"included","exclusionReason":null,"notes":["Technical evidence retained for batch provenance; not a separate leaderboard row."]}]}],"batches":[{"id":"adaptive-planning-batch-1","categoryId":"workday-adaptive-planning","title":"Adaptive Planning · Batch 1","shortTitle":"Batch 1","summary":"Model quality on 21 Workday Adaptive Planning questions, using each model’s first completed answer to each question.","datasetVersion":"0.1.0-draft","rubricVersion":"0.1.0-draft","publishedAt":"2026-09-28T18:21:27.601Z","questionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"qualityReviewStatus":"provisional-ai-reviewed","reviewMethod":"Provisional blinded AI-assisted rubric review. Documentation-backed credit includes calibrated partial credit without treating undocumented alternatives as proven false. Focused cross-answer calibration covered archived versions, prior-period recurrence and import modes. No independent expert or tenant validation was performed.","coverageNote":"All 21 models completed and received provisional assessments for all 21 questions. Every score uses the first completed answer per question. All 532 attempts, including 91 noncompletions, remain in completion and cost accounting.","limitations":["Scores are provisional documentation-backed rubric results, not independently validated ERP accuracy or statistically established rankings.","The first completed answer is selected per model and question, never the highest-scoring retry. Original, recovery and completion requests used different limits; this is not an equal-compute comparison.","Latency is the median of selected completed-answer elapsed times. Costs cover all candidate API attempts, including failures; unknown charges keep total cost unknown. Assistant review and infrastructure costs are not measured.","Unverified behavior receives no unsupported credit under the owner-approved rule; this does not prove an undocumented alternative is false.","Original collection requested 8192 output tokens and a 180-second timeout. The first 61 attempts used socket-inactivity timeout enforcement; the remaining 359 used a wall-clock deadline. This is not a uniform-deadline cohort.","Failure-selected recovery only: originally completed answers were not repeated. Expanded 32768-token and 600-second request limits, with paced requests; historical owner stops and noncompletion cutoffs remain in the attempt evidence.","Supplemental model collection used 32768-token and 600-second limits. A narrow owner-approved retention exception applied only to these nonconfidential benchmark prompts; provider retention was possible, training was disallowed, and no tools or credentials were supplied.","Missing-answer completion only; earlier completed answers are retained. Requests used 32768-token and 600-second limits, fixed original provider routes, ZDR and no-training restrictions. Three consecutive noncompletions paused dispatch for diagnosis; no model was silently excluded.","The final 17 missing MiMo answers used owner-approved Novita after DeepInfra rate limits, with the same model and frozen questions. Requests retained ZDR, no training, no fallback, 32768-token and 600-second limits. FP8 was catalog metadata, not independently verified inference precision.","Quality scores use the first completed answer per model/question; incomplete assessment coverage is unscored."],"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-01-2026-09","adaptive-planning-recovery-02-2026-09","adaptive-planning-recovery-03-2026-09","adaptive-planning-recovery-04-2026-09","adaptive-planning-recovery-05-2026-09","adaptive-planning-recovery-06-2026-09","adaptive-planning-recovery-07-2026-09","adaptive-planning-recovery-08-2026-09","adaptive-planning-recovery-09-2026-09","adaptive-planning-recovery-10-2026-09","adaptive-planning-recovery-11-2026-09","adaptive-planning-recovery-12-2026-09","adaptive-planning-recovery-13-2026-09","adaptive-planning-supplement-2026-09","adaptive-planning-completion-deepseek-v41-flash-2026-09","adaptive-planning-completion-hy4-2026-09","adaptive-planning-completion-mimo-v25-2026-09","adaptive-planning-completion-mimo-novita-2026-09"],"models":[{"modelId":"deepseek/deepseek-v4.1-flash","modelName":"DeepSeek: DeepSeek V4.1 Flash","provider":"InferenceNet","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":74.52380952380952,"medianLatencyMs":227426,"attemptCount":25,"statusCounts":{"completed":21,"truncated":1,"api-error":2,"transport-error":0,"protocol-error":1,"empty":0,"interrupted":0},"knownCostUsd":0.025421215,"unknownCostAttemptCount":2,"totalCostUsd":null,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-11-2026-09","adaptive-planning-completion-deepseek-v41-flash-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"z-ai/glm-5.3-flash","modelName":"Z.ai: GLM 5.3 Flash","provider":"Z.AI","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":65,"medianLatencyMs":51685,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.01594333,"unknownCostAttemptCount":0,"totalCostUsd":0.01594333,"sourceRunIds":["adaptive-planning-original-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"tencent/hy4-preview","modelName":"Tencent: Hy4 preview","provider":"Tencent","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":72.61904761904762,"medianLatencyMs":408313,"attemptCount":43,"statusCounts":{"completed":21,"truncated":17,"api-error":0,"transport-error":2,"protocol-error":1,"empty":0,"interrupted":2},"knownCostUsd":1.102974089,"unknownCostAttemptCount":2,"totalCostUsd":null,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-05-2026-09","adaptive-planning-completion-hy4-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"openai/gpt-5.6-luna","modelName":"OpenAI: GPT-5.6 Luna","provider":"Azure","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":73.0952380952381,"medianLatencyMs":16148,"attemptCount":22,"statusCounts":{"completed":21,"truncated":1,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.0849202,"unknownCostAttemptCount":0,"totalCostUsd":0.0849202,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-07-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"deepseek/deepseek-v4-flash-0731","modelName":"DeepSeek: DeepSeek V4 Flash 0731","provider":"Makora","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":66.42857142857143,"medianLatencyMs":10877,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.00958899,"unknownCostAttemptCount":0,"totalCostUsd":0.00958899,"sourceRunIds":["adaptive-planning-original-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"xiaomi/mimo-v2.5","modelName":"Xiaomi: MiMo-V2.5","provider":"DeepInfra, Novita","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":55.95238095238095,"medianLatencyMs":33339,"attemptCount":50,"statusCounts":{"completed":21,"truncated":0,"api-error":27,"transport-error":2,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.014363881,"unknownCostAttemptCount":27,"totalCostUsd":null,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-12-2026-09","adaptive-planning-completion-mimo-v25-2026-09","adaptive-planning-completion-mimo-novita-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"deepseek/deepseek-v4-flash","modelName":"DeepSeek: DeepSeek V4 Flash 0423","provider":"Venice","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":65.71428571428571,"medianLatencyMs":29676,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.004762954,"unknownCostAttemptCount":0,"totalCostUsd":0.004762954,"sourceRunIds":["adaptive-planning-original-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"tencent/hy3","modelName":"Tencent: Hy3","provider":"Tencent","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":72.14285714285714,"medianLatencyMs":59185,"attemptCount":29,"statusCounts":{"completed":21,"truncated":8,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.0759984225,"unknownCostAttemptCount":0,"totalCostUsd":0.0759984225,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-09-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"z-ai/glm-5.3","modelName":"Z.ai: GLM 5.3","provider":"Z.AI","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":74.28571428571429,"medianLatencyMs":18863,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.12607444,"unknownCostAttemptCount":0,"totalCostUsd":0.12607444,"sourceRunIds":["adaptive-planning-original-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"google/gemini-3.8-flash","modelName":"Google: Gemini 3.8 Flash","provider":"Google","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":77.85714285714286,"medianLatencyMs":15304,"attemptCount":23,"statusCounts":{"completed":21,"truncated":2,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.2784915,"unknownCostAttemptCount":0,"totalCostUsd":0.2784915,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-04-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"openai/gpt-6-luna","modelName":"OpenAI: GPT-6 Luna","provider":"Azure","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":74.04761904761905,"medianLatencyMs":13446,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.0261795,"unknownCostAttemptCount":0,"totalCostUsd":0.0261795,"sourceRunIds":["adaptive-planning-original-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"upstage/solar-pro4","modelName":"Upstage: Solar Pro 4","provider":"Upstage","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":66.19047619047619,"medianLatencyMs":28261,"attemptCount":22,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":1,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.01963107,"unknownCostAttemptCount":0,"totalCostUsd":0.01963107,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-10-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"z-ai/glm-5.2","modelName":"Z.ai: GLM 5.2","provider":"Z.AI","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":73.0952380952381,"medianLatencyMs":26575,"attemptCount":26,"statusCounts":{"completed":21,"truncated":5,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.51820272,"unknownCostAttemptCount":0,"totalCostUsd":0.51820272,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-03-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"openai/gpt-5.6-sol","modelName":"OpenAI: GPT-5.6 Sol","provider":"Azure","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":83.57142857142857,"medianLatencyMs":15202,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.56622,"unknownCostAttemptCount":0,"totalCostUsd":0.56622,"sourceRunIds":["adaptive-planning-original-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"anthropic/claude-sonnet-5","modelName":"Anthropic: Claude Sonnet 5","provider":"Amazon Bedrock","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":66.9047619047619,"medianLatencyMs":11922,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.247086,"unknownCostAttemptCount":0,"totalCostUsd":0.247086,"sourceRunIds":["adaptive-planning-original-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"minimax/minimax-m3","modelName":"MiniMax: MiniMax M3","provider":"Together","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":62.857142857142854,"medianLatencyMs":10445,"attemptCount":22,"statusCounts":{"completed":21,"truncated":1,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.04637676,"unknownCostAttemptCount":0,"totalCostUsd":0.04637676,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-08-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"moonshotai/kimi-k3","modelName":"MoonshotAI: Kimi K3","provider":"Moonshot AI","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":87.14285714285714,"medianLatencyMs":61583,"attemptCount":23,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":2,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0.722394,"unknownCostAttemptCount":0,"totalCostUsd":0.722394,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-02-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"inclusionai/ling-3.0-flash-fin:free","modelName":"inclusionAI: Ling 3.0 Flash Fin (free)","provider":"Novita","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":66.42857142857143,"medianLatencyMs":4612,"attemptCount":33,"statusCounts":{"completed":21,"truncated":0,"api-error":12,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0,"unknownCostAttemptCount":12,"totalCostUsd":null,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-13-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"openai/gpt-6-astra","modelName":"OpenAI: GPT-6 Astra","provider":"Azure","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":82.85714285714286,"medianLatencyMs":20665,"attemptCount":22,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":1,"empty":0,"interrupted":0},"knownCostUsd":1.41055,"unknownCostAttemptCount":1,"totalCostUsd":null,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-01-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"deepseek/deepseek-v4-pro-0813","modelName":"DeepSeek: DeepSeek V4 Pro 0813","provider":"Ionstream","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":73.80952380952381,"medianLatencyMs":53777,"attemptCount":24,"statusCounts":{"completed":21,"truncated":2,"api-error":0,"transport-error":0,"protocol-error":0,"empty":1,"interrupted":0},"knownCostUsd":0.222388944,"unknownCostAttemptCount":0,"totalCostUsd":0.222388944,"sourceRunIds":["adaptive-planning-original-2026-09","adaptive-planning-recovery-06-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."},{"modelId":"stealth/space-bunny-alpha","modelName":"Space Bunny Alpha","provider":"Stealth","completedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"assessedQuestionIds":["AP-F01","AP-F02","AP-F03","AP-F04","AP-F05","AP-F06","AP-F07","AP-F08","AP-F09","AP-F10","AP-C01","AP-C02","AP-C03","AP-C04","AP-C05","AP-C06","AP-C07","AP-C08","AP-C09","AP-C10","AP-I01"],"heldQuestionIds":[],"qualityScore":66.42857142857143,"medianLatencyMs":34114,"attemptCount":21,"statusCounts":{"completed":21,"truncated":0,"api-error":0,"transport-error":0,"protocol-error":0,"empty":0,"interrupted":0},"knownCostUsd":0,"unknownCostAttemptCount":0,"totalCostUsd":0,"sourceRunIds":["adaptive-planning-supplement-2026-09"],"notes":"Score uses the first completed answer per question; attempt totals include every source attempt."}]}]}