Across 6 shared benchmarks, Claude Sonnet 4-20250514 (no thinking) scores higher on 3 and o4-mini on 3. The widest gap is vectara_hallucination_rate, where Claude Sonnet 4-20250514 (no thinking) scores 10.3 against 18.6.
| Benchmark | Claude Sonnet 4-20250514 (no thinking) | o4-mini |
|---|---|---|
| aider_polyglot | 56.4 | 72 |
| arena_vision | 1176 | 1201 |
| vectara_answer_rate | 98.6 | 99.2 |
| vectara_avg_summary_length | 145.8 | 130.9 |
| vectara_factual_consistency | 89.7 | 81.4 |
| vectara_hallucination_rate ↓ | 10.3 | 18.6 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.