OpenAI
GPT-5.2 benchmark results
35 results on 11 benchmarks from 6 sources.
Also listed as: gpt-5.2-high · gpt-5.2 · gpt-5.2-2025-12-11_none · GPT-5.2 (none) · gpt-5.2-2025-12-11_xhigh · GPT-5.2 (xhigh) · gpt-5.2-2025-12-11_high · GPT-5.2 (high) · gpt-5.2-2025-12-11_medium · GPT-5.2 (medium)
All benchmark results
| Benchmarks | Score | Rank | Setting | Source type | Date |
|---|---|---|---|---|---|
| LMArena Text | 1437 | 101 | Overall, style control on; 49685 votes; 95% CI 1434–1441“gpt-5.2-high” | Official leaderboard | Oct 8, 2026 |
| LMArena Text | 1436 | 104 | Overall, style control on; 83928 votes; 95% CI 1432–1439“gpt-5.2” | Official leaderboard | Oct 8, 2026 |
| LMArena WebDev | 1416 | 83 | Overall; 1082 votes; 95% CI 1394–1438“gpt-5.2” | Official leaderboard | Oct 8, 2026 |
| LMArena Vision | 1243 | 57 | Overall, style control on; 19039 votes; 95% CI 1237–1249“gpt-5.2-high” | Official leaderboard | Oct 2, 2026 |
| LMArena Vision | 1229 | 68 | Overall, style control on; 20134 votes; 95% CI 1223–1235“gpt-5.2” | Official leaderboard | Oct 2, 2026 |
| GPQA Diamond | 91.4% | 29 | Epoch AI run, accuracy; effort: xhigh | Independent run | Dec 13, 2025 |
| GPQA Diamond | 88.2% | 60 | Epoch AI run, accuracy; effort: high | Independent run | Dec 11, 2025 |
| GPQA Diamond | 87.9% | 62 | Epoch AI run, accuracy; effort: medium | Independent run | Dec 11, 2025 |
| GPQA Diamond | 82.7% | 102 | Epoch AI run, accuracy; effort: low | Independent run | Dec 11, 2025 |
| GPQA Diamond | 73.2% | 148 | Epoch AI run, accuracy; effort: none | Independent run | Jul 13, 2026 |
| MMLU-Pro | 87.4% | 10 | Overall accuracy; data source: Self-Reported | Vendor-reported | Mar 11, 2026 |
| SWE-bench Verified | 73.8% | 17 | Epoch AI run (Epoch scaffold); effort: high | Independent run | Feb 12, 2026 |
| SWE-bench Verified | 72.8% | 16 | agent: mini-SWE-agent“gpt-5-2” | Official leaderboard | Feb 17, 2026 |
| AIME (OTIS Mock AIME 2024–2025) | 96.1% | 40 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: high | Independent run | Dec 11, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 96.1% | 42 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: xhigh | Independent run | Dec 13, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 93.9% | 54 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: medium | Independent run | Dec 11, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 78.9% | 113 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: low | Independent run | Dec 11, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 62.2% | 144 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: none | Independent run | Jul 13, 2026 |
| Humanity's Last Exam | 27.8% | 14 | Compiled by Epoch AI; effort: unknown thinking | Third-party compilation via Epoch AI | Dec 11, 2025 |
| τ²-bench | 83.0% | 2 | airline domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | May 5, 2026 |
| τ²-bench | 52.5% | 22 | airline domain, pass^1; effort: none; submitted by Sierra; v1.0.1 | Official leaderboard | May 26, 2026 |
| τ²-bench | 32.2% | 13 | banking_knowledge domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | May 5, 2026 |
| τ²-bench | 12.6% | 26 | banking_knowledge domain, pass^1; effort: none; submitted by Sierra; v1.0.1 | Official leaderboard | May 26, 2026 |
| τ²-bench | 81.6% | 7 | retail domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | May 5, 2026 |
| τ²-bench | 75.0% | 17 | retail domain, pass^1; effort: none; submitted by Sierra; v1.0.1 | Official leaderboard | May 26, 2026 |
| τ²-bench | 89.7% | 11 | telecom domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | May 5, 2026 |
| τ²-bench | 57.2% | 17 | telecom domain, pass^1; effort: none; submitted by Sierra; v1.0.1 | Official leaderboard | May 26, 2026 |
| Terminal-Bench 2.0 | 64.9% | 33 | Terminal-Bench 2.0; agent: Droid; effort: unknown thinking | Official leaderboard via Epoch AI | Dec 11, 2025 |
| Terminal-Bench 2.0 | 64.9% | 34 | Terminal-Bench 2.0; agent: Droid; effort: medium | Official leaderboard via Epoch AI | Dec 24, 2025 |
| Terminal-Bench 2.0 | 62.9% | 39 | Terminal-Bench 2.0; agent: Codex CLI; effort: unknown thinking | Official leaderboard via Epoch AI | Dec 11, 2025 |
| Terminal-Bench 2.0 | 62.9% | 41 | Terminal-Bench 2.0; agent: Codex CLI; effort: medium | Official leaderboard via Epoch AI | Dec 18, 2025 |
| Terminal-Bench 2.0 | 60.7% | 47 | Terminal-Bench 2.0; agent: Mux; effort: unknown thinking | Official leaderboard via Epoch AI | Dec 11, 2025 |
| Terminal-Bench 2.0 | 54.0% | 61 | Terminal-Bench 2.0; agent: Terminus 2; effort: unknown thinking | Official leaderboard via Epoch AI | Dec 11, 2025 |
| BFCL v4 (function calling) | 55.9% | 16 | Overall accuracy; native function calling (FC)“GPT-5.2-2025-12-11 (FC)” | Official leaderboard | Oct 17, 2026 |
| BFCL v4 (function calling) | 45.3% | 38 | Overall accuracy; prompt-based“GPT-5.2-2025-12-11 (Prompt)” | Official leaderboard | Oct 17, 2026 |
FAQ
Where do these GPT-5.2 scores come from?
From Epoch AI Benchmarking Hub, Berkeley Function Calling Leaderboard (BFCL), 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.