GPT-5 (medium)
Unknown SizeBy OpenAI • Released 2025-08-07
Capability Radar
Avg Score
65
Across all benchmarks
Participated
14
Benchmarks
Benchmark Performance
| Benchmark | Category | Score |
|---|---|---|
| MATH-500 | Reasoning | 99.1 |
| AIME 2025 | Reasoning | 91.7 |
| MMLU-Pro | Knowledge | 86.7 |
| 𝜏²-Bench Telecom | Reasoning Knowledge | 86.5 |
| GPQA Diamond | Knowledge | 84.2 |
| LCR | Long-Context Reasoning | 72.8 |
| IFBench | Agent | 70.6 |
| LiveCodeBench | Coding | 70.3 |
| SWE-bench (Bash Only) | Coding Agent | 65 |
| Artificial Analysis Intelligence Index | Knowledge | 41.8 |
| SciCode | Reasoning Knowledge | 41.1 |
| Artificial Analysis Coding Index | Coding | 39 |
| Terminal-Bench Hard | Agent Coding | 37.9 |
| HLE | Knowledge Multi-Modal | 23.5 |