GPT-5 Pro
GPT-5 Pro ranks first on DeepSearchQA.
27
Benchmarks with a score
1
Ranked first, out of five or more models
2
In the top three
OpenAI Oct 2025
GPT-5 Pro
Closed weights
151.2
#63 of 256
Epoch Capabilities Index
- Humanity's Last Exam 31.6% #12 /47
- FrontierMath Tier 4 14.6% #13 /55
- PRBench 49.9% #11 /48
Where it ranks
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.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MultiNRC | 65.2% | #2 of 38 | Official leaderboard | |
| Humanity's Last Exam | 31.6% | #12 of 47 | Official leaderboard | |
| ELI5 Arena | 1,463 | #10 of 18 | Official leaderboard | |
| Joke Arena | 1,488 | #17 of 25 | Official leaderboard | |
| Corp Speak Arena | 1,460 | #15 of 22 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| FrontierMath Tier 4 | 14.6% | #13 of 55 | Epoch AI run | |
| SimpleBench | 61.6% | #37 of 106 | Official leaderboard | |
| IMO-ProofBench | 28.6% | #7 of 20 | Official leaderboard | |
| Putnam 2025 | 93.0 | #2 of 4 | Official leaderboard | |
| FrontierMath Tiers 1-3 v2 | 55.8% | #47 of 86 | Official leaderboard | |
| ARC-AGI-2 | 18.3% | #54 of 98 | Official leaderboard | |
| ARC-AGI | 70.2% | #54 of 95 | Official leaderboard | |
| FrontierMath Tier 4 v2 | 19.5% | #48 of 68 | Official leaderboard | |
| Miklós Schweitzer 2025 | 90% | #1 of 1 | Official leaderboard |
Coding
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DeepSearchQA | 65.2% | #1 of 10 | Official leaderboard | |
| Chess Puzzles | 48% | #19 of 93 | Epoch AI run | |
| xbench-DeepSearch-2510 | 79% | #1 of 2 | Official leaderboard |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PRBench | 49.9% | #11 of 48 | Official leaderboard |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| VISTA | 52.4% | #4 of 54 | Official leaderboard | |
| ClockBench | 11.1% | #28 of 37 | Official leaderboard | |
| VoxelBench | 1,199 | #56 of 56 | Official leaderboard |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| FORTRESS | 15.2% | #16 of 73 | Official leaderboard | |
| Human Pathogen Capabilities Test | 55.9% | #34 of 86 | Official leaderboard | |
| Molecular Biology Capabilities Test | 49.6% | #41 of 90 | Official leaderboard | |
| World-Class Bio | 43.6% | #42 of 74 | Official leaderboard |
OpenAI's models over time
OpenAI's models in color, against every other model on the Epoch Capabilities Index, by release date.
Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)