GPT-5vsClaude Opus 4
Across 11 shared benchmarks, GPT-5 leads overall: GPT-5 wins 10, Claude Opus 4 wins 1, with 0 ties and an average score difference of +16.18.
GPT-5
OpenAI · 2025-08-07 · Foundation model
Claude Opus 4
Anthropic · 2025-05-23 · Reasoning model
GPT-510 wins(91%)(9%)1 winClaude Opus 4
Benchmark scores
Grouped by capability, sorted by largest gap within each. 11 shared benchmarks.
Math and Reasoning
GPT-5 4/5| Benchmark | GPT-5 | Claude Opus 4 | Diff |
|---|---|---|---|
| IMO-ProofBench | 592 / 16 | 2.9016 / 16 | +56.10 |
| AIME2025 | 99.609 / 106 | 75.5065 / 106 | +24.10 |
| FrontierMath | 24.8015 / 60 | 4.5039 / 60 | +20.30 |
| FrontierMath - Tier 4 | 12.5029 / 80Thinking High (No Tools) | 4.2040 / 80 | +8.30 |
| Simple Bench | 56.708 / 27 | 58.807 / 27 | -2.10 |
General Knowledge
GPT-5 4/4| Benchmark | GPT-5 | Claude Opus 4 | Diff |
|---|---|---|---|
| ARC-AGI | 65.7030 / 65 | 35.7048 / 65 | +30 |
| HLE | 35.2060 / 157 | 10.70129 / 157 | +24.50 |
| GPQA Diamond | 87.3037 / 178 | 79.6079 / 178 | +7.70 |
| ARC-AGI-2 | 9.9037 / 59 | 8.6039 / 59 | +1.30 |
Agent Level Benchmark
GPT-5 1/1| Benchmark | GPT-5 | Claude Opus 4 | Diff |
|---|---|---|---|
| τ²-Bench | 8015 / 40 | 72.5022 / 40 | +7.50 |
Coding and Software Engineer
GPT-5 1/1| Benchmark | GPT-5 | Claude Opus 4 | Diff |
|---|---|---|---|
| SWE-bench Verified | 72.8046 / 108 | 72.5048 / 108 | +0.30 |
Specs
| Field | GPT-5 | Claude Opus 4 |
|---|---|---|
| Publisher | OpenAI | Anthropic |
| Release date | 2025-08-07 | 2025-05-23 |
| Model type | Foundation model | Reasoning model |
| Architecture | Dense | Dense |
| Parameters | Not available | Not available |
| Context length | 400K | 200K |
| Max output | 128K | 32K |
Summary
- GPT-5leads in:Math and Reasoning (4/5), General Knowledge (4/4), Agent Level Benchmark (1/1), Coding and Software Engineer (1/1)
On average across the 11 shared benchmarks, GPT-5 scores 16.18 higher.
Largest single-benchmark gap: IMO-ProofBench — GPT-5 59 vs Claude Opus 4 2.90 (+56.10).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.