GPT-5.2
OpenAI·Released Dec 11, 2025
Updated Oct 8, 9:18 PM ET
Overall score
48.4
Not ranked yet: fewer than 3 kinds of coding evaluation
Evidence
19 results
52% of scored kinds covered
Context window
400K tokens
From OpenRouter
API price, USD per 1M tokens
$1.75 / $14
Input / output · cached input $0.18
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.
Coding
48.6
2 results · 22% coverage
Research and reasoning
52.6
6 results · 63% coverage
Professional work
52.0
4 results · 33% coverage
Knowledge and accuracy
40.5
6 results · 100% coverage
Human preference
40.1
1 result · 100% coverage
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
2 results · 48.6- LiveBench CodingLiveBench · high effort63.2%Reading 49.3
- Vibe Code BenchVals AI · xhigh effort53.5%Reading 48.6
Research and reasoning
6 results · 52.6- Chess PuzzlesEpoch AI · high effort40.0%Reading 60.0
- LiveBench ReasoningLiveBench · high effort88.2%Reading 57.1
- FrontierMath Tiers 1–3Epoch AI · xhigh effort67.4%Reading 55.0
- Humanity's Last Exam (AA run)Artificial Analysis · xhigh effort37.7%Reading 53.5
- FrontierMath Tier 4Epoch AI · xhigh effort31.7%Reading 51.3
- Mystery Game PuzzlesEpoch AI · high effort23.0%Reading 42.2
Professional work
4 results · 52.0- LiveBench Data AnalysisLiveBench · high effort78.2%Reading 68.0
- MedCodeVals AI · xhigh effort49.7%Reading 67.0
- τ²-Bench Telecom (AA run)Artificial Analysis · xhigh effort84.8%Reading 36.0
- τ-Bench Banking (AA run)Artificial Analysis · none effort11.1%Reading 11.2
Knowledge and accuracy
6 results · 40.5- LiveBench LanguageLiveBench · high effort79.8%Reading 57.3
- AA-LCRArtificial Analysis · xhigh effort82.7%Reading 53.9
- IFBench (AA run)Artificial Analysis · xhigh effort75.4%Reading 53.6
- SimpleQA VerifiedEpoch AI · high effort34.3%Reading 31.8
- LiveBench Instruction FollowingLiveBench · high effort61.8%Reading 30.5
- BullshitBenchBullshitBench · high effortCapped23.6%Reading 6.9
Human preference
1 result · 40.1- Arena TextLMArena · high effort1,437.3Reading 36.7
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.
- LiveCodeBenchVals AI · CodingReference only85.4%
- LiveCodeBench (AA run)Artificial Analysis · CodingReference only88.9%
- SWE-bench VerifiedEpoch AI · CodingReference only73.8%
- SWE-bench VerifiedVals AI · CodingReference only75.8%
- AIME (AA run)Artificial Analysis · Research and reasoningReference only99.0%
- OTIS Mock AIMEEpoch AI · Research and reasoningReference only96.1%
- CorpFinVals AI · Professional workReference only65.9%
- EBR-benchEpoch AI · Professional workWatching23.0%
- LegalBenchVals AI · Professional workReference only82.8%
- MedScribeVals AI · Professional workWatching84.4%
- TaxEvalVals AI · Professional workReference only75.8%
- GPQA DiamondEpoch AI · Knowledge and accuracyReference only88.2%
- GPQA DiamondVals AI · Knowledge and accuracyReference only91.7%
- MMLU-ProVals AI · Knowledge and accuracyReference only86.2%
- AA Intelligence IndexArtificial Analysis · Composite indicesReference only30.4
- LiveBench averageLiveBench · Composite indicesReference only74.6%
- Arena VisionLMArena · VisionReference only1,242.8
- MMMU ProVals AI · VisionReference only86.7%
- Arena WebDevLMArena · Writing and designReference only1,416.1
- MGSMVals AI · MultilingualReference only94.0%
No result yet
Scored evaluations this model has not taken. A missing result neither adds nor subtracts.
- Terminal-Bench 4.0 · Vals AI
- Terminal-Bench 4.0 (AA run) · Artificial Analysis
- Code Migration · Vals AI
- ProgramBench · Vals AI
- Vibe Code Bench 1–100 · Vals AI
- CyberBench Patch · Vals AI
- APEX-SWE · Mercor
- SciCode · Artificial Analysis
- MysteryMechanism · Vals AI
- Terminal-Bench Science · Vals AI
- ProofBench · Vals AI
- Finance Agent · Vals AI
- APEX-Agents · Mercor
- Harvey Legal Agent Benchmark · Vals AI
- Legal Research Bench · Vals AI
- Tax Agent Bench · Vals AI
- EMB · Vals AI