GPT-5.6 Terra
OpenAI·Released Jul 9, 2026
Updated Oct 8, 11:55 PM ET
Overall score
56.3
Rank #27 overall
Evidence
35 results
97% of scored kinds covered
Context window
1.1M tokens
From OpenRouter
API price, USD per 1M tokens
$2 / $12
Input / output · cached input $0.20
The score is on the board's scale (reference average 50, about 15 points per standard deviation), not a percentage.
By domain
Each domain is scored on the same scale as the overall score. Coverage is the share of the domain's kinds of evaluation with a result; it says how much evidence there is, not how good the model is.
Every result behind the score
The result as the evaluator publishes it, and the reading it gives on the board's scale. A hard evaluation can give a high reading for a modest result; results near 95% or 5% only say "at least" or "at most".
Coding
10 results · 57.1- CyberBench PatchVals AI · max effort87.5%Reading 75.2
- Code MigrationVals AI · max effort47.8%Reading 66.4
- ProgramBenchVals AI · max effortNear the floor0.5%Reading at most 65.7
- Vibe Code BenchVals AI · max effort74.6%Reading 60.1
- Terminal-Bench 4.0Vals AI · max effort22.7%Reading 59.2
- LiveBench CodingLiveBench · max effort66.6%Reading 58.2
- SciCodeArtificial Analysis · high effort52.4%Reading 56.1
- Vibe Code Bench 1–100Vals AI · max effort14.8%Reading 53.8
- APEX-SWEMercor · max effort31.1%Reading 38.5
- Terminal-Bench 4.0 (AA run)Artificial Analysis · high effortNear the floor1.5%Reading at most 34.5
Research and reasoning
8 results · 65.5- Chess PuzzlesEpoch AI · max effort54.0%Reading 77.0
- FrontierMath Tiers 1–3Epoch AI · max effort86.0%Reading 74.4
- LiveBench ReasoningLiveBench · max effort92.8%Reading 74.0
- FrontierMath Tier 4Epoch AI · max effort70.7%Reading 73.4
- ProofBenchVals AI · max effort74.0%Reading 67.8
- Terminal-Bench ScienceVals AI · max effort10.0%Reading 58.7
- Mystery Game PuzzlesEpoch AI · max effort35.0%Reading 56.8
- Humanity's Last Exam (AA run)Artificial Analysis · high effort38.5%Reading 54.6
Professional work
10 results · 55.6- LiveBench Data AnalysisLiveBench · max effort79.3%Reading 72.0
- APEX-AgentsMercor · max effort58.2%Reading 66.8
- EMBVals AI · max effort66.2%Reading 64.2
- Legal Research BenchVals AI · max effort41.3%Reading 63.5
- Tax Agent BenchVals AI · max effort30.0%Reading 63.1
- Finance AgentVals AI · max effort54.4%Reading 62.4
- τ-Bench Banking (AA run)Artificial Analysis · high effort28.7%Reading 47.3
- MedCodeVals AI · xhigh effort43.4%Reading 45.7
- τ²-Bench Telecom (AA run)Artificial Analysis · high effort78.4%Reading 27.6
- Harvey Legal Agent BenchmarkVals AI · max effortNear the floor0.8%Reading at most 24.8
Knowledge and accuracy
6 results · 45.7- LiveBench LanguageLiveBench · max effort82.9%Reading 66.6
- AA-LCRArtificial Analysis · high effort77.7%Reading 45.3
- SimpleQA VerifiedEpoch AI · max effort43.2%Reading 45.0
- LiveBench Instruction FollowingLiveBench · max effort64.6%Reading 41.2
- BullshitBenchBullshitBench · max effort34.5%Reading 41.0
- IFBench (AA run)Artificial Analysis · high effort64.4%Reading 29.8
Human preference
1 result · 48.7- Arena TextLMArena · xhigh effort1,465.5Reading 45.1
Shown, not scored
Composite indices, saturated or older evaluations, evaluations still being checked, and vision, writing and multilingual results. They do not move the score.
- IOIVals AI · CodingReference only87.6%
- LiveCodeBenchVals AI · CodingReference only85.9%
- SkillsBenchVals AI · CodingWatching58.9%
- SRE BenchVals AI · CodingWatching9.5%
- SWE-bench VerifiedVals AI · CodingReference only95.4%
- OTIS Mock AIMEEpoch AI · Research and reasoningReference only99.7%
- APEX-AccountingMercor · Professional workWatching6.0%
- CorpFinVals AI · Professional workReference only65.3%
- LegalBenchVals AI · Professional workReference only85.1%
- MedScribeVals AI · Professional workWatching82.9%
- TaxEvalVals AI · Professional workReference only76.2%
- GPQA DiamondEpoch AI · Knowledge and accuracyReference only93.3%
- GPQA DiamondVals AI · Knowledge and accuracyReference only90.9%
- MMLU-ProVals AI · Knowledge and accuracyReference only86.7%
- AA Intelligence IndexArtificial Analysis · Composite indicesReference only34.2
- LiveBench averageLiveBench · Composite indicesReference only77.9%
- Vals IndexVals AI · Composite indicesReference only53.1%
- Arena VisionLMArena · VisionReference only1,268.2
- MMMU ProVals AI · VisionReference only86.5%
No result yet
Scored evaluations this model has not taken. A missing result neither adds nor subtracts.
- MysteryMechanism · Vals AI