OpenAI
availableShows if the model has enough results for an index.GPT-5.4
GPT-5.4 is a reasoning model from OpenAI. 62 benchmarks count toward its score, in 7 categories.
IndexOverall score out of 100.67.6 ±2.6
CoverageShare of the index weight with results.95%
SpeedOutput tokens per second.32/s
Input / 1MUS dollars per 1M input tokens.$2.5 batch $1.25
Output / 1MUS dollars per 1M output tokens.$15 batch $7.5 US dollars per 1M output tokens in a batch.
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1453 (#41)
The index is a score out of 100. The ± range shows how much it can change. Batch work costs less.
63,526 votes. Elo shows what people prefer. It does not change the score.
CapabilitiesScore per category, out of 100.
Out of 100Results
62 counted| BenchmarkThe test name. | CategoryThe capability that the test measures. | ResultThe score from the publisher. | IndexThis result as a score out of 100. | RunThe settings of the run. | DateDate of the result. | Published byThe source of the result. |
|---|---|---|---|---|---|---|
| τ²-Bench Tool-Agent-User Evaluation | Agentic | 98.9% | 69.8 | — | — | Victor Barres et al. |
| AIME | Math | 96.7% | 58.8 | — | 16 Apr 2026 | Vals AI |
| OTIS Mock AIME 2024-2025 | Math | 95.3% | 64.8 | xhigh effort | — | Epoch AI |
| LiveBench Mathematics | Math | 94.1% | 73.7 | xhigh effort | 25 Jun 2026 | LiveBench |
| ARC-AGI-1 (semi-private) | Reasoning | 93.7% | 71.4 | xhigh effort | — | ARC Prize Foundation |
| GPQA diamond | Knowledge | 93.3% | 64.4 | xhigh effort | — | Epoch AI |
| Graduate-Level Google-Proof Q&A | Knowledge | 92.8% | 63.9 | — | — | David Rein et al. |
| GPQA Diamond | Knowledge | 92.8% | 63.9 | — | — | David Rein et al. |
| Artificial Analysis GPQA Diamond | Knowledge | 92.0% | 62.8 | — | — | Artificial Analysis |
| GPQA Diamond | Knowledge | 91.7% | 62.9 | — | 1 Sept 2026 | Vals AI |
| LiveBench Reasoning | Reasoning | 88.1% | 78.4 | xhigh effort | 25 Jun 2026 | LiveBench |
| MMMU Pro | Multimodal | 87.5% | 69.8 | — | 1 Sept 2026 | Vals AI |
| LiveCodeBench Pro | Coding | 87.5% | — | — | — | LiveCodeBench Pro authors |
| MMLU Pro | Knowledge | 87.5% | 58.4 | — | 1 Sept 2026 | Vals AI |
| ScreenSpot Pro | Multimodal | 85.4% | 67.5 | — | — | Kaixin Li et al. |
| React Native Evals | Coding | 85.3% | 66.8 | — | — | Callstack |
| LiveCodeBench | Coding | 84.1% | 60.8 | — | 1 Sept 2026 | Vals AI |
| CharXiv Reasoning | Multimodal | 82.8% | 60.4 | — | — | CharXiv authors |
| BrowseComp | Agentic | 82.7% | 69.2 | — | — | OpenAI |
| LiveBench Language | Knowledge | 82.6% | 70.2 | xhigh effort | 25 Jun 2026 | LiveBench |
| MMMU-Pro with Python | Multimodal | 82.1% | — | — | — | OpenAI |
| Artificial Analysis Long Context Reasoning | Reasoning | 82.0% | 65.0 | — | — | Artificial Analysis |
| Massive Multi-discipline Multimodal Understanding Pro | Multimodal | 81.2% | 59.5 | — | — | MMMU-Pro authors |
| LiveBench Data Analysis | Reasoning | 79.3% | 66.1 | xhigh effort | 25 Jun 2026 | LiveBench |
| CyberGym | Agentic | 79.0% | 70.2 | — | — | Zhun Wang et al. |
| FrontierMath-Tiers-1-3-v2-Private | Math | 78.6% | 75.7 | xhigh effort | — | Epoch AI |
| Artificial Analysis MMMU-Pro | Multimodal | 78.4% | 62.9 | — | — | Artificial Analysis |
| SWE-bench | Coding | 78.2% | 60.3 | — | 1 Sept 2026 | Vals AI |
| LiveBench Coding | Coding | 77.5% | 66.4 | xhigh effort | 25 Jun 2026 | LiveBench |
| MedXpertQA Multimodal | Multimodal | 77.1% | — | — | — | Meta AI |
| SWE-Bench verified | Coding | 76.9% | 59.2 | high effort | — | Epoch AI |
| OSWorld-Verified | Agentic | 75.0% | 63.6 | — | — | Tianbao Xie et al. |
| ARC-AGI-2 (semi-private) | Reasoning | 74.0% | 77.0 | xhigh effort | — | ARC Prize Foundation |
| Artificial Analysis IFBench | Instruction | 73.9% | 64.1 | — | — | Artificial Analysis |
| DeepSearchQA | Agentic | 73.6% | 56.5 | — | — | Meta AI |
| Artificial Analysis Coding Index | Coding | 71.0% | 69.0 | — | — | Artificial Analysis |
| MCP Atlas | Agentic | 70.6% | 61.3 | — | — | OpenAI |
| LiveBench Instruction Following | Instruction | 70.2% | 70.8 | xhigh effort | 25 Jun 2026 | LiveBench |
| IOI v1 | Coding | 67.8% | 78.5 | — | 9 Aug 2026 | Vals AI |
| ERQA | Multimodal | 65.4% | 60.0 | — | — | Qwen |
| Gert Labs Composite Game Benchmark | Agentic | 64.9% | 70.0 | — | — | Gert Labs |
| SimpleVQA | Multimodal | 61.1% | 51.8 | — | — | Z.AI |
| Claw-Eval | Agentic | 60.3% | 53.2 | — | — | Bowen Ye et al. |
| MedXpertQA Text | Knowledge | 59.6% | — | — | — | Meta AI |
| Terminal-Bench 2.0 | Agentic | 58.4% | 64.4 | — | 4 Jun 2026 | Vals AI |
| Vibe Code Bench v1.1 | Coding | 57.9% | 66.3 | OpenHands | 21 Sept 2026 | Vals AI |
| SWE-bench Pro | Coding | 57.7% | 59.8 | — | — | Xiang Deng et al. |
| Toolathlon | Agentic | 54.6% | 68.9 | — | — | OpenAI |
| LiveBench Agentic Coding | Agentic | 53.8% | 67.4 | xhigh effort | 25 Jun 2026 | LiveBench |
| OfficeQA Pro | Multimodal | 53.2% | 65.5 | — | — | OfficeQA Pro authors |
| Humanity's Last Exam | Knowledge | 52.1% | 72.9 | — | — | Center for AI Safety et al. |
| SkillsBench | Coding | 51.7% | 66.8 | OpenHands | 11 Sept 2026 | Vals AI |
| Artificial Analysis Omniscience Accuracy | Knowledge | 50.8% | 76.7 | — | — | Artificial Analysis |
| FrontierMath-Tier-4-v2-Private | Math | 49.0% | 71.5 | xhigh effort | — | Epoch AI |
| HealthBench Professional | Knowledge | 48.1% | — | — | — | Rebecca Soskin Hicks et al. |
| FrontierMath-2025-02-28-Private | Math | 47.6% | 75.0 | xhigh effort | — | Epoch AI |
| SimpleQA Verified | Knowledge | 45.1% | 63.3 | xhigh effort | — | Epoch AI |
| Chess Puzzles | Reasoning | 44.0% | 80.4 | xhigh effort | — | Epoch AI |
| Artificial Analysis Humanity's Last Exam | Knowledge | 43.7% | 71.9 | — | — | Artificial Analysis |
| ZeroBench | Multimodal | 41.0% | — | — | — | Meta AI |
| HealthBench Hard | Knowledge | 40.1% | 82.8 | — | — | Meta AI |
| Humanity's Last Exam without tools | Knowledge | 39.8% | 62.5 | — | — | OpenAI |
| τ²-bench Banking | Agentic | 39.4% | 27.1 | xhigh effort · Sierra | 4 Aug 2026 | Sierra Research |
| Artificial Analysis Intelligence Index | Knowledge | 39.0% | 71.1 | — | — | Artificial Analysis |
| JobBench | Agentic | 38.9% | 60.9 | — | — | Yuetai Li et al. |
| Furniture Assembly | Reasoning | 37.5% | 66.2 | xhigh effort | — | Epoch AI |
| Mystery Game Puzzles | Reasoning | 37.0% | 73.3 | xhigh effort | — | Epoch AI |
| GDPval-AA normalized | Agentic | 36.6% | 63.7 | — | — | Artificial Analysis |
| Code Migration | Coding | 35.0% | 67.3 | — | 21 Sept 2026 | Vals AI |
| APEX-Agents-AA | Agentic | 33.3% | 67.4 | — | — | Artificial Analysis / Mercor |
| FrontierMath-Tier-4-2025-07-01-Private | Math | 27.1% | 72.9 | xhigh effort | — | Epoch AI |
| EBR-bench | Reasoning | 25.4% | 66.7 | xhigh effort | — | Epoch AI |
| Critical Physics Tasks | Reasoning | 23.4% | 87.4 | — | — | Artificial Analysis |
| MirrorCode | Coding | 15.6% | — | high effort | — | Epoch AI |
| ResearchClawBench | Agentic | 15.3% | — | — | — | InternScience |
| ApprenticeBench: end-to-end computer use, continual learning, and long-horizon agency on a real accounts-payable job | Agentic | 11.0% | 66.2 | — | — | NeoCognition |
| ExploitGym | Agentic | 6.0% | 65.6 | — | — | Zhun Wang et al. |
| Agent Arena command recovery | Agentic | 4.6 | 72.6 | high effort | 15 Sept 2026 | LMArena |
| Agent Arena steerability | Agentic | 2.5 | 70.2 | high effort | 15 Sept 2026 | LMArena |
| ProgramBench | Coding | 0.5% | — | — | 21 Sept 2026 | Vals AI |
| ARC-AGI-3 (semi-private) | Reasoning | 0.2% | — | high effort | — | ARC Prize Foundation |
| ProgramBench | Coding | 0.0% | — | — | 21 Sept 2026 | Vals AI |
| Agent Arena task outcome | Agentic | -1.6 | 65.5 | high effort | 15 Sept 2026 | LMArena |
62 benchmarks count, from 72 of 83 results. A grey row does not count. Too few models took that benchmark.