GPT-5.5 ranks first on 26 of the 242 benchmarks it has scores on.
Closed weights Flagship
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Wer wird Millionär? LLM Benchmark | 588889.0 | #2 of 76 | Official leaderboard | |
| Judgemark v4 | 87.8 | #2 of 40 | Official leaderboard | |
| Extended NYT Connections | 96.2% | #6 of 97 | Official leaderboard | |
| CL-bench Life | 22.2% | #1 of 16 | Official leaderboard | |
| FACTS Search | 85% | #2 of 31 | Official leaderboard | |
| LMArena (Elo) | 1,482 | #23 of 333 | Official leaderboard | |
| SEA-HELM | 76.3 | #6 of 62 | Official leaderboard | |
| EQ-Bench 3 | 1,554 | #8 of 78 | Official leaderboard | |
| FACTS Parametric | 78% | #6 of 45 | Official leaderboard | |
| Creative Writing v3 | 79.5 | #17 of 120 | Official leaderboard | |
| SimpleQA Verified | 63% | #14 of 84 | Epoch AI | |
| Longform Creative Writing (EQ-Bench) | 78.2 | #21 of 119 | Official leaderboard | |
| EQ-Bench 4 | 1,295 | #6 of 32 | Official leaderboard | |
| FACTS Multimodal | 45.8% | #8 of 39 | Official leaderboard | |
| Document Arena | 1,483 | #8 of 37 | Official leaderboard | |
| Phare | 70.5% | #19 of 62 | Official leaderboard | |
| BullshitBench V2 | 47% | #50 of 122 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 9.30% | #52 of 104 | Official leaderboard | |
| FACTS Benchmark Suite | 45.8% | #20 of 40 | Official leaderboard | |
| BullshitBench | 27.3% | #54 of 107 | Official leaderboard | |
| OpenAI user-flagged hallucination eval | 9.20% | #4 of 4 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 100% | #1 of 202 | Epoch AI | |
| FrontierMath | 51.7% | #2 of 70 | Epoch AI | |
| AIME 2026 | 100% | #1 of 31 | Official leaderboard | |
| HMMT 2026 | 98.5% | #1 of 31 | Official leaderboard | |
| MathArena (Apex, IMO, USAMO) | 80.2% | #2 of 47 | Official leaderboard | |
| GPQA Diamond | 94% | #12 of 219 | Epoch AI | |
| FrontierMath Tier 4 | 35.4% | #4 of 55 | Epoch AI | |
| DTBench | 93.3% | #14 of 178 | Official leaderboard | |
| ResearchClawBench | 34.9 | #2 of 23 | Official leaderboard | |
| CritPt | 27.1% | #16 of 160 | Epoch AI | |
| USAMO 2026 | 98.2% | #1 of 9 | Official leaderboard | |
| ARC-AGI-2 | 85% | #11 of 95 | Official leaderboard | |
| Surface Evolver Bench | 88.1% | #4 of 26 | Epoch AI | |
| FrontierMath Tiers 1-3 v2 | 85.3% | #14 of 89 | Epoch AI | |
| ARC-AGI | 95% | #17 of 92 | Official leaderboard | |
| FrontierMath Tier 4 v2 | 72.5% | #14 of 71 | Epoch AI | |
| BTF-3 | 0.13 | #2 of 10 | Epoch AI | |
| SimpleBench | 69% | #21 of 100 | Official leaderboard | |
| Verifying the Verifiers (Project APE CRED benchmark) | 96.5% | #5 of 23 | Official leaderboard | |
| ProofBench | 50% | #13 of 59 | Official leaderboard | |
| IMO-ProofBench | 71.9% | #3 of 10 | Official leaderboard | |
| Leipzig Benchmark | 44% | #9 of 19 | Official leaderboard | |
| FrontierMath Erdős | 0% | #7 of 14 | Epoch AI | |
| BixBench3 | 43.1 | #11 of 20 | Official leaderboard | |
| LatchBio Capabilities v1.0 | 41.9% | #15 of 27 | Official leaderboard | |
| ForecastBench | 64.2 | #65 of 86 | Official leaderboard | |
| NatureBench | 4.44% | #17 of 18 | Official leaderboard | |
| BixBench | 80.5% | #4 of 4 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| WeirdML | 84.9% | #7 of 112 | Official leaderboard | |
| SWE-bench Verified | 80.6% | #2 of 32 | Epoch AI | |
| SlopCodeBench | 28.1% | #1 of 14 | Official leaderboard | |
| Bird-SQL | 72.5% | #1 of 13 | Official leaderboard | |
| Monthly-SWEBench | 76% | #1 of 12 | Official leaderboard | |
| ALE-Bench | 1,943 | #12 of 121 | Epoch AI | |
| LiveSQLBench | 33.5% | #5 of 36 | Official leaderboard | |
| OpenHands Index | 65.9% | #5 of 33 | Official leaderboard | |
| EvoCode-Bench | 52.4% | #2 of 13 | Official leaderboard | |
| EvoCode-Bench v2 | 29.5% | #2 of 13 | Official leaderboard | |
| FrontierCode | 25.5% | #2 of 12 | Official leaderboard | |
| FrontierCS | 72.1 | #3 of 14 | Official leaderboard | |
| PostTrainBench | 28.4% | #5 of 23 | Official leaderboard | |
| Code Migration | 45.2% | #18 of 78 | Official leaderboard | |
| AndroidBench | 80.2% | #8 of 32 | Official leaderboard | |
| GSO-Bench | 40.2% | #8 of 30 | Official leaderboard | |
| GBAEval | 53.2% | #6 of 21 | Official leaderboard | |
| Senior SWE-Bench | 29.5% | #8 of 23 | Official leaderboard | |
| SWE Atlas Refactoring | 44.8% | #6 of 17 | Official leaderboard | |
| ProgramBench | 0.50% | #9 of 25 | Official leaderboard | |
| SWE Atlas | 45.4% | #8 of 22 | Official leaderboard | |
| FrontierCode 1.1 | 43% | #3 of 8 | Official leaderboard | |
| SWE-bench-Live | 62% | #5 of 13 | Official leaderboard | |
| MLE-bench | 36.7% | #7 of 18 | Model card | |
| WebDev Arena | 1,510 | #41 of 105 | Epoch AI | |
| DeepSWE | 67% | #11 of 28 | Official leaderboard | |
| Vibe Code Bench | 69.8% | #46 of 115 | Official leaderboard | |
| Monorepo-Bench | 60% | #2 of 5 | Model card | |
| SWE Atlas Test Writing | 42.6% | #9 of 22 | Official leaderboard | |
| Code Arena | 1,512 | #53 of 124 | Official leaderboard | |
| TasteVal | 47.0× | #9 of 20 | Official leaderboard | |
| SWE-Marathon | 12% | #6 of 12 | Official leaderboard | |
| FrontierCode 2.0 | 22.6% | #15 of 26 | Official leaderboard | |
| Image-to-WebDev Arena | 1,532 | #32 of 55 | Official leaderboard | |
| MLS-Bench | 35.5% | #10 of 15 | Official leaderboard | |
| PostTrainBench 1.1 | 27.2% | #8 of 11 | Official leaderboard | |
| PostTrainBench 1.2 | 23.9% | #14 of 17 | Official leaderboard | |
| RSI-Exam | 33.1% | #15 of 18 | Official leaderboard | |
| Internal Research Debugging Evaluation | 50.5% | #5 of 6 | Model card | |
| Vals RSI Index | 0.17 | #21 of 25 | Official leaderboard | |
| OpenAI-Proof Q&A | 1.70% | #8 of 9 | Model card | |
| GameDevBench | 54.7% | #9 of 10 | Official leaderboard | |
| MirrorCode | 10% | #26 of 28 | Epoch AI |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GAIA | 91.7% | #1 of 57 | Official leaderboard | |
| SnakeBench | 37 | #4 of 183 | Official leaderboard | |
| Terminal-Bench 2.0 | 84.7% | #1 of 43 | Official leaderboard | |
| Game Arena Bargaining | 277 | #1 of 36 | Official leaderboard | |
| Game Arena Coin Game | 514 | #1 of 35 | Official leaderboard | |
| tau2-bench | 98% | #2 of 69 | Model card | |
| Pencil Puzzle Bench | 83.3% | #2 of 53 | Official leaderboard | |
| PACT | 1,607 | #1 of 26 | Official leaderboard | |
| Toolathlon | 55.6% | #2 of 50 | Official leaderboard | |
| SkillsBench | 67.3% | #1 of 21 | Official leaderboard | |
| Game Arena Dark Hex | 482 | #2 of 38 | Official leaderboard | |
| Game Arena Word Association | 552 | #2 of 36 | Official leaderboard | |
| LLM Chess | 1,532 | #8 of 124 | Official leaderboard | |
| Game Arena Poker | 25.3 | #2 of 31 | Official leaderboard | |
| Chess Puzzles | 63% | #6 of 83 | Epoch AI | |
| Game Arena Five in a Row | 925 | #2 of 20 | Official leaderboard | |
| BrowseComp | 90.1% | #8 of 77 | Model card | |
| LisanBench | 9844.7 | #14 of 133 | Official leaderboard | |
| Harbor-Index | 28.1% | #1 of 9 | Official leaderboard | |
| Mystery Game Puzzles | 56% | #10 of 80 | Epoch AI | |
| Game Arena Checkers | 1,130 | #5 of 38 | Official leaderboard | |
| Game Arena Reversi | 790 | #6 of 36 | Official leaderboard | |
| Game Arena Ultimate Tic-Tac-Toe | 486 | #6 of 36 | Official leaderboard | |
| DeepResearch Bench | 54% | #4 of 24 | Epoch AI | |
| LiveClawBench | 74.5% | #3 of 17 | Official leaderboard | |
| Game Arena Chess | 1,257 | #7 of 39 | Official leaderboard | |
| Kaggle Game Arena | 407 | #7 of 39 | Official leaderboard | |
| Game Arena Lines of Action | 714 | #7 of 36 | Official leaderboard | |
| τ³-bench Banking | 44.6% | #5 of 25 | Official leaderboard | |
| MobileWorld | 62.4% | #4 of 18 | Official leaderboard | |
| LMArena Search Arena | 1,226 | #8 of 35 | Official leaderboard | |
| Terminal-Bench 2.1 | 83.2% | #3 of 13 | Official leaderboard | |
| Game Arena Dots and Boxes | 990 | #10 of 37 | Official leaderboard | |
| Game Arena Clobber | 410 | #10 of 36 | Official leaderboard | |
| Toolathlon-Verified | 73.5% | #7 of 25 | Official leaderboard | |
| Game Arena Word Art | 1,021 | #10 of 35 | Official leaderboard | |
| ClawBench V2 | 35.4% | #2 of 7 | Official leaderboard | |
| Game Arena Four in a Row | 476 | #12 of 38 | Official leaderboard | |
| OSWorld-Verified | 78.7% | #12 of 37 | Model card | |
| Long-Horizon-Terminal-Bench (LHTB) | 44.5% | #9 of 25 | Official leaderboard | |
| ApprenticeBench | 20% | #7 of 18 | Official leaderboard | |
| EdgeBench | 48.4 | #2 of 5 | Official leaderboard | |
| PinchBench 2.0 | 75.5% | #24 of 57 | Official leaderboard | |
| EBR-bench | 34.3% | #12 of 28 | Epoch AI | |
| OSWorld 2.0 | 13% | #3 of 7 | Official leaderboard | |
| DeepTerminalBench | 32% | #4 of 9 | Official leaderboard | |
| Game Arena Go | 998 | #17 of 36 | Official leaderboard | |
| Game Arena Nine Men's Morris | 620 | #17 of 36 | Official leaderboard | |
| STATE-Bench | 58.9% | #1 of 2 | Official leaderboard | |
| MCP Atlas | 75.3% | #19 of 34 | Official leaderboard | |
| MyPCBench | 45.1% | #6 of 9 | Official leaderboard | |
| ARC-AGI-3 | 0.43% | #17 of 24 | Official leaderboard | |
| CEO-Bench | $33,260 | #14 of 19 | Official leaderboard | |
| Time Horizon Index: KSP | 8.33% | #10 of 13 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| FORTE | 77.8% | #1 of 21 | Official leaderboard | |
| CorpFin v2 | 68.4% | #9 of 119 | Official leaderboard | |
| ProfBench | 58% | #6 of 72 | Official leaderboard | |
| LegalBench | 86.5% | #13 of 138 | Official leaderboard | |
| EnterpriseClawBench | 66.3% | #1 of 9 | Official leaderboard | |
| MortgageTax | 68.8% | #10 of 86 | Official leaderboard | |
| Vals Finance Agent | 60% | #6 of 49 | Official leaderboard | |
| Vending-Bench Arena | $10,627 | #4 of 32 | Official leaderboard | |
| CaseLaw v2 | 66.2% | #7 of 53 | Official leaderboard | |
| prinzbench | 74.0 | #4 of 26 | Official leaderboard | |
| PhysicianBench | 46.3% | #2 of 13 | Official leaderboard | |
| E-Commerce Bench | 7.02 | #3 of 19 | Official leaderboard | |
| TaxEval v2 | 75% | #25 of 129 | Official leaderboard | |
| GDPval | 84.9% | #2 of 10 | Model card | |
| MedScribe | 86.9% | #20 of 98 | Official leaderboard | |
| RedlineBench | 50.5% | #1 of 4 | Official leaderboard | |
| TaxCalcBench v2 | 64% | #7 of 26 | Official leaderboard | |
| Vending-Bench 2 | $7,524 | #18 of 66 | Official leaderboard | |
| PolicyBench | 88% | #14 of 50 | Official leaderboard | |
| MedCode | 49.1% | #27 of 96 | Official leaderboard | |
| Excel Modeling Benchmark | 64.5% | #26 of 76 | Official leaderboard | |
| Legal Research Bench | 40.4% | #31 of 88 | Official leaderboard | |
| CADGenBench | 36% | #4 of 11 | Official leaderboard | |
| APEX-Agents | 55.1% | #21 of 56 | Epoch AI | |
| FoodTruck Bench | $61,408 | #4 of 10 | Official leaderboard | |
| HealthBench Consensus | 95.6% | #7 of 17 | Model card | |
| Remote Labor Index | 6.25% | #7 of 17 | Official leaderboard | |
| Vals Finance Agent v2 | 51.8% | #39 of 80 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 3.75% | #40 of 80 | Official leaderboard | |
| CommerceAgentBench | 47.7% | #7 of 13 | Official leaderboard | |
| Tax Agent Bench | 21.8% | #46 of 79 | Official leaderboard | |
| JobBench | 38.3% | #10 of 16 | Official leaderboard | |
| Public Benefits Bench | 60.9% | #32 of 50 | Official leaderboard | |
| HealthBench Hard | 31.5% | #14 of 21 | Model card | |
| FrontierFinance | 43.5% | #12 of 16 | Official leaderboard | |
| HealthBench | 56.5% | #23 of 30 | Model card | |
| HealthBench Professional | 51.8% | #23 of 25 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Math Kangaroo 2025 (MathArena) | 94.9% | #1 of 22 | Official leaderboard | |
| RoboRewardBench | 0.66 | #1 of 19 | Official leaderboard | |
| Code2Video Benchmark | 1,571 | #2 of 19 | Official leaderboard | |
| Vision Arena | 1,286 | #16 of 136 | Official leaderboard | |
| ZeroBench | 23% | #8 of 66 | Official leaderboard | |
| BabyVision | 83.6% | #5 of 27 | Official leaderboard | |
| VoxelBench | 1,939 | #14 of 56 | Official leaderboard | |
| ClockBench | 46.1% | #11 of 37 | Official leaderboard | |
| GDP.pdf | 26% | #12 of 39 | Epoch AI | |
| Blueprint-Bench 2 | 36.2% | #11 of 35 | Official leaderboard | |
| Furniture Assembly Benchmark | 44.2% | #14 of 35 | Epoch AI | |
| ParseBench | 66.6% | #20 of 42 | Official leaderboard | |
| Drone-Bench | 63.9% | #10 of 19 | Official leaderboard | |
| LongExtractionBench | 52.7% | #2 of 3 | Official leaderboard | |
| ArtifactArena | 967 | #17 of 21 | Official leaderboard | |
| RoboDojo | 1.13% | #5 of 6 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Capture the Flag (Professional) | 96.3% | #1 of 21 | Model card | |
| Cyber Range | 93.3% | #1 of 10 | Model card | |
| First-Person Fairness evaluation | 1.12% | #1 of 10 | Model card | |
| CyberBench (Vals) | 80.5% | #3 of 26 | Official leaderboard | |
| KORA | 67% | #5 of 42 | Official leaderboard | |
| Destructive action avoidance | 90% | #1 of 8 | Model card | |
| Human Pathogen Capabilities Test v2 (HPCT v2) | 66.2% | #9 of 71 | Official leaderboard | |
| SEC-bench Pro | 46.4% | #1 of 6 | Official leaderboard | |
| SEC-bench Pro 260617 | 58.4% | #1 of 6 | Official leaderboard | |
| EpiBench | 28.3% | #5 of 27 | Official leaderboard | |
| BrowseSafe++ | 89.3% | #3 of 15 | Official leaderboard | |
| CVE-Bench | 93.1% | #3 of 14 | Model card | |
| ExploitBench | 72% | #2 of 9 | Official leaderboard | |
| FORTRESS | 16.3% | #16 of 66 | Official leaderboard | |
| BiomniBench | 63.6 | #2 of 8 | Official leaderboard | |
| Tacit Knowledge and Troubleshooting | 78.3% | #8 of 29 | Model card | |
| SpatialBench-Long | 32.1% | #8 of 27 | Official leaderboard | |
| CyberGym | 81.8% | #10 of 33 | Official leaderboard | |
| TroubleshootingBench | 44.1% | #6 of 19 | Model card | |
| PCT (Polarized Contrastive Pairs) | 40% | #16 of 46 | Official leaderboard | |
| ProtocolQA Open-Ended | 37.3% | #9 of 25 | Model card | |
| DrugDiscoveryBench | 51.6% | #7 of 19 | Official leaderboard | |
| scBench-Long | 21.6% | #10 of 27 | Official leaderboard | |
| Multimodal Troubleshooting Virology | 53.7% | #11 of 26 | Model card | |
| DeepsecBench | 21.0 | #12 of 28 | Official leaderboard | |
| Dictatorship Eval | 82.5% | #9 of 20 | Official leaderboard | |
| TxBench-Preclinical-Pharmacology | 55.3% | #13 of 28 | Official leaderboard | |
| BioSecBench-Surveillance | 47.1% | #13 of 27 | Official leaderboard | |
| CyScenarioBench | 26% | #7 of 14 | Model card | |
| DNA sequence design for transcription factor binding | 13.8% | #2 of 4 | Model card | |
| Hard-negative protein binding prediction | 1.48% | #2 of 4 | Model card | |
| GeneBench | 33.2% | #1 of 2 | Model card | |
| SRE-Bench | 3.82% | #17 of 33 | Official leaderboard | |
| ExploitGym | 14.8% | #10 of 18 | Official leaderboard | |
| TxBench-Oligo-Discovery | 33.9% | #17 of 28 | Official leaderboard | |
| OncologyBench | 14.7% | #19 of 30 | Official leaderboard | |
| TxBench-Antibody-Discovery | 33% | #21 of 28 | Official leaderboard | |
| UK AISI cyber ranges | 10% | #3 of 4 | Model card | |
| VariantBench | 30.5% | #21 of 27 | Official leaderboard | |
| BioSecBench-Function | 28.3% | #22 of 27 | Official leaderboard | |
| CoT-Control | 0.20% | #8 of 8 | Model card | |
| Agent JSK | 96.3% | #7 of 7 | Model card | |
| OpenAI Deployment Simulation | 0.01% | #5 of 5 | Model card | |
| Biochemistry knowledge improvement | 32.3% | #2 of 2 | Model card | |
| UK AISI narrow cyber tasks | 90.5% | #2 of 2 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| ACCoRD | 80.5 | #10 of 150 | Official leaderboard | |
| Conceptual Reasoning Index | 67.3 | #14 of 149 | Official leaderboard | |
| LMCA | 54.3% | #16 of 139 | Epoch AI | |
| SAGE (Vals) | 51.5% | #13 of 81 | Official leaderboard | |
| CAIS Text Capabilities Index | 46.3 | #13 of 66 | Official leaderboard | |
| Artificial Analysis Intelligence Index v4.0 | 60.2 | #1 of 5 | Model card | |
| CAIS Vision Capabilities Index | 64.3 | #13 of 53 | Official leaderboard | |
| Vals Multimodal Index | 68.1% | #9 of 33 | Official leaderboard | |
| CAIS Risk Index | 42.4 | #18 of 65 | Official leaderboard |
OpenAI's models in color, against every other model on the Epoch Capabilities Index, by release date.
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it. 40 of them aren't in the launch post.