OpenAI
GPT-5 benchmark results
33 results on 11 benchmarks from 5 sources.
Also listed as: gpt-5-high · gpt-5-chat · gpt-5-medium · gpt-5-2025-08-07_minimal · GPT-5 (minimal) · gpt-5-2025-08-07_high · GPT-5 (high) · gpt-5-2025-08-07_medium · GPT-5 (medium) · gpt-5-2025-08-07
All benchmark results
| Benchmarks | Score | Rank | Setting | Source type | Date |
|---|---|---|---|---|---|
| LMArena Text | 1435 | 105 | Overall, style control on; 31723 votes; 95% CI 1430–1439“gpt-5-high” | Official leaderboard | Oct 8, 2026 |
| LMArena Text | 1426 | 118 | Overall, style control on; 31203 votes; 95% CI 1422–1431“gpt-5-chat” | Official leaderboard | Oct 8, 2026 |
| LMArena WebDev | 1418 | 82 | Overall; 3092 votes; 95% CI 1402–1433“gpt-5-medium” | Official leaderboard | Oct 8, 2026 |
| LMArena Vision | 1226 | 70 | Overall, style control on; 39910 votes; 95% CI 1219–1233“gpt-5-chat” | Official leaderboard | Oct 2, 2026 |
| LMArena Vision | 1210 | 79 | Overall, style control on; 34456 votes; 95% CI 1203–1217“gpt-5-high” | Official leaderboard | Oct 2, 2026 |
| GPQA Diamond | 86.2% | 75 | Epoch AI run, accuracy; effort: high | Independent run | Oct 29, 2025 |
| GPQA Diamond | 85.3% | 82 | Epoch AI run, accuracy; effort: medium | Independent run | Aug 7, 2025 |
| GPQA Diamond | 71.7% | 153 | Epoch AI run, accuracy; effort: minimal | Independent run | Jul 20, 2026 |
| MMLU-Pro | 87.1% | 16 | Overall accuracy; data source: Self-Reported“GPT-5(high)” | Vendor-reported | Mar 11, 2026 |
| SWE-bench Verified | 73.5% | 19 | Epoch AI run (Epoch scaffold); effort: high | Independent run | Feb 6, 2026 |
| SWE-bench Verified | 71.5% | 23 | Epoch AI run (Epoch scaffold); effort: medium | Independent run | Feb 5, 2026 |
| SWE-bench Verified | 74.4% | 12 | agent: Prometheus-v1.2.1“gpt-5-2025-08-07” | Official leaderboard | Oct 15, 2025 |
| SWE-bench Verified | 71.8% | 20 | agent: OpenHands“openai/gpt-5-2025-08-07” | Official leaderboard | Aug 7, 2025 |
| SWE-bench Verified | 71.2% | 22 | agent: Prometheus-v1.2“gpt-5-2025-08-07” | Official leaderboard | Sep 29, 2025 |
| SWE-bench Verified | 65.0% | 38 | agent: mini-SWE-agent“gpt-5-2025-08-07” | Official leaderboard | Aug 7, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 91.4% | 65 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: high | Independent run | Oct 29, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 87.2% | 82 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: medium | Independent run | Aug 7, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 46.7% | 172 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: minimal | Independent run | Jul 20, 2026 |
| MATH Level 5 | 98.1% | 1 | Epoch AI run, accuracy; effort: high | Independent run | Oct 29, 2025 |
| MATH Level 5 | 97.9% | 2 | Epoch AI run, accuracy; effort: medium | Independent run | Aug 20, 2025 |
| Humanity's Last Exam | 25.3% | 15 | Compiled by Epoch AI; effort: high | Third-party compilation via Epoch AI | Aug 7, 2025 |
| Humanity's Last Exam | 25.3% | 15 | Compiled by Epoch AI; effort: unknown thinking | Third-party compilation via Epoch AI | Aug 7, 2025 |
| τ²-bench | 62.5% | 14 | airline domain, pass^1; submitted by Sierra; v0.1.3 | Official leaderboard | Aug 9, 2025 |
| τ²-bench | 81.6% | 7 | retail domain, pass^1; submitted by Sierra; v0.1.3 | Official leaderboard | Aug 9, 2025 |
| τ²-bench | 95.8% | 6 | telecom domain, pass^1; submitted by Sierra; v0.1.3 | Official leaderboard | Aug 9, 2025 |
| Terminal-Bench 2.0 | 49.6% | 70 | Terminal-Bench 2.0; agent: Codex CLI; effort: unknown thinking | Official leaderboard via Epoch AI | Aug 7, 2025 |
| Terminal-Bench 2.0 | 49.6% | 71 | Terminal-Bench 2.0; agent: Codex CLI ; effort: medium | Official leaderboard via Epoch AI | Nov 4, 2025 |
| Terminal-Bench 2.0 | 43.8% | 80 | Terminal-Bench 2.0; agent: OpenHands; effort: unknown thinking | Official leaderboard via Epoch AI | Aug 7, 2025 |
| Terminal-Bench 2.0 | 43.8% | 81 | Terminal-Bench 2.0; agent: OpenHands ; effort: medium | Official leaderboard via Epoch AI | Nov 2, 2025 |
| Terminal-Bench 2.0 | 35.2% | 113 | Terminal-Bench 2.0; agent: Terminus 2 ; effort: medium | Official leaderboard via Epoch AI | Oct 31, 2025 |
| Terminal-Bench 2.0 | 35.2% | 114 | Terminal-Bench 2.0; agent: Terminus 2; effort: unknown thinking | Official leaderboard via Epoch AI | Aug 7, 2025 |
| Terminal-Bench 2.0 | 33.9% | 120 | Terminal-Bench 2.0; agent: Mini-SWE-Agent; effort: unknown thinking | Official leaderboard via Epoch AI | Aug 7, 2025 |
| Terminal-Bench 2.0 | 33.9% | 121 | Terminal-Bench 2.0; agent: Mini-SWE-Agent ; effort: medium | Official leaderboard via Epoch AI | Nov 3, 2025 |
FAQ
Where do these GPT-5 scores come from?
From Epoch AI Benchmarking Hub, LMArena leaderboard dataset, MMLU-Pro leaderboard (TIGER-Lab), SWE-bench leaderboards (swebench.com), and τ²-bench leaderboard submissions. Each row links to the original leaderboard, dataset or report, and vendor-reported numbers are labelled.