New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
GPT-5
About GPT-5
GPT-5 is an AI model developed by OpenAI, in the multimodal category, released in 2025, made available as a proprietary (API-only) model.
On this page you'll find GPT-5's full specifications, including 75 benchmark results. Review provider pricing and benchmark scores below, or compare GPT-5 head-to-head with other multimodal models.
Links
Community
No reports yet. Own this model on some hardware? Be the first to confirm it runs.
No comments yet. Start the discussion.
GPT-5 benchmark scores▶
| Benchmark | Category | Score | Variant |
|---|---|---|---|
| BrowseComp | agentic | 54.9% | cited-deepseek-v3-2 |
| BrowseComp-Zh | agentic | 63 | cited-deepseek-v3-2 |
| MCP-Mark | agentic | 50.9 | cited-deepseek-v3-2 |
| MCP-Universe | agentic | 47.9 | cited-deepseek-v3-2 |
| Terminal-Bench 2.0 | agentic | 35.2% | cited-deepseek-v3-2 |
| Tool-Decathlon | agentic | 29 | cited-deepseek-v3-2 |
| τ²-Bench (TAU2) | agentic | 80.2 | cited-deepseek-v3-2 |
| Codeforces Elo | code | 2,537 | rating-cited-deepseek-v3-2 |
| Codeforces Elo | code | 1,918 | thinking-high-rating-cited-ring-1t |
| Codeforces Elo | code | 1,120 | main-rating-cited-ling-1t |
| FullStackBench | code | 50.92% | main-cited-ling-1t |
| LiveCodeBench | code | 84.5% | cited-deepseek-v3-2 |
| LiveCodeBench | code | 58.57% | main-cited-ling-1t |
| LiveCodeBench v5 | code | 82.68% | High |
| LiveCodeBench v6 | code | 80.6% | thinking-high-cited-ring-1t |
| MBPP | code | 91.72% | main-sanitized-cited-ling-1t |
| SWE-bench Verified | code | 74.9% | cited-deepseek-v3-2 |
| Aider Benchmark | coding | 95.49 | thinking-high-cited-ring-1t |
| Aider Benchmark | coding | 84.4 | main-edit-cited-ling-1t |
| MultiPL-E | coding | 76.66 | main-cited-ling-1t |
| SWE-Bench Multilingual | coding | 55.3 | cited-deepseek-v3-2 |
| ArenaHard v2 | general | 82.91 | thinking-high-win-rate-cited-ring-1t |
| ArenaHard v2 | general | 65.06 | main-win-rate-cited-ling-1t |
| C-Eval | general | 88.27% | thinking-high-cited-ring-1t |
| C-Eval | general | 83.59% | main-cited-ling-1t |
| IFEval | general | 95.38% | thinking-high-cited-ring-1t |
| IFEval | general | 85.11% | main-prompt-strict-cited-ling-1t |
| MMLU-Pro | general | 87.5% | cited-deepseek-v3-2 |
| MMLU-Pro | general | 86.21% | thinking-high-cited-ring-1t |
| MMLU-Pro | general | 81.94% | main-cited-ling-1t |
| MMLU-Redux | general | 92.75% | main-cited-ling-1t |
| MultiChallenge | general | 69.6% | thinking-high-cited-ring-1t |
| MultiChallenge | general | 48.72% | main-cited-ling-1t |
| AGIEval | knowledge | 88.56 | thinking-high-cited-ring-1t |
| CMMLU | knowledge | 87.09 | thinking-high-cited-ring-1t |
| TriviaQA | knowledge | 86.32 | thinking-high-cited-ring-1t |
| AIME 2024 | math | 94.78% | High |
| AIME 2024 | math | 67.6% | main-cited-ling-1t |
| AIME 2025 | math | 94.6% | cited-deepseek-v3-2 |
| AIME 2025 | math | 94.6% | thinking-high-cited-ring-1t |
| AIME 2025 | math | 92.15% | High |
| AIME 2025 | math | 59.43% | main-cited-ling-1t |
| CNMO 2024 | math | 87.93% | thinking-high-cited-ring-1t |
| CNMO 2024 | math | 63.11% | main-cited-ling-1t |
| HMMT February 2025 | math | 93.3% | thinking-high-cited-ring-1t |
| HMMT February 2025 | math | 91.79% | High |
| HMMT February 2025 | math | 88.3% | cited-deepseek-v3-2 |
| HMMT February 2025 | math | 36.98% | main-cited-ling-1t |
| HMMT November 2025 | math | 89.2 | cited-deepseek-v3-2 |
| IMOAnswerBench | math | 76 | cited-deepseek-v3-2 |
| Omni-MATH | math | 82.9 | thinking-high-cited-ring-1t |
| Omni-MATH | math | 61.09 | main-cited-ling-1t |
| Omni-MATH-HARD | math | 73.61% | High |
| UGMathBench | math | 80.18 | thinking-high-cited-ring-1t |
| UGMathBench | math | 67.27 | main-cited-ling-1t |
| ARC-AGI-1 | reasoning | 65.7 | thinking-high-cited-ring-1t |
| ARC-AGI-1 | reasoning | 14.06 | main-cited-ling-1t |
| BBEH | reasoning | 72.78 | thinking-high-cited-ring-1t |
| BBEH | reasoning | 39.75 | main-cited-ling-1t |
| FinanceReasoning | reasoning | 89.33 | thinking-high-cited-ring-1t |
| FinanceReasoning | reasoning | 86.28 | main-cited-ling-1t |
| GPQA Diamond | reasoning | 86.05% | thinking-high-cited-ring-1t |
| GPQA Diamond | reasoning | 85.7% | cited-deepseek-v3-2 |
| GPQA Diamond | reasoning | 71.31% | main-cited-ling-1t |
| Humanity's Last Exam | reasoning | 35.2% | search-agent-cited-deepseek-v3-2 |
| Humanity's Last Exam | reasoning | 28.63% | High |
| Humanity's Last Exam | reasoning | 26.3% | text-only-cited-deepseek-v3-2 |
| Humanity's Last Exam | reasoning | 24.84% | thinking-high-cited-ring-1t |
| Humanity's Last Exam | reasoning | 7.33% | main-cited-ling-1t |
| ZebraLogic | reasoning | 98% | thinking-high-cited-ring-1t |
| ZebraLogic | reasoning | 57.3% | main-cited-ling-1t |
| HealthBench | safety | 67.2% | thinking-high-cited-ring-1t |
| PhyBench | science | 48.53 | thinking-high-cited-ring-1t |
| BFCL v3 | tool-use | 57.21% | thinking-high-cited-ring-1t |
| BFCL v3 | tool-use | 50.27% | main-cited-ling-1t |
More models like GPT-5
Frequently asked questions
What is GPT-5? ▶
GPT-5 is an AI model developed by OpenAI, in the multimodal category, released in 2025. It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.
Who created GPT-5? ▶
GPT-5 was developed by OpenAI and released in 2025.
Is GPT-5 open source or proprietary? ▶
GPT-5 is a proprietary model. It is accessed through an API rather than by downloading the weights.
How much does GPT-5 cost? ▶
Pricing for GPT-5 depends on the provider. See the providers table on this page for the latest API rates.
How does GPT-5 perform on benchmarks? ▶
GPT-5 is benchmarked across 75 evaluations on GenAIList, including BrowseComp (54.9%). See the full benchmark table below and compare it with other models.
No videos yet. Know a good one? Add it below.
Know a great video about GPT-5?
Add YouTube Video
Add Tweet/Discussion
Report an Issue
Help us maintain accurate information in our database.
Suggest an Edit
Tell us what needs to be changed or corrected for