GPT-5 mini scores 18.3 on the Respan Index (#103), ahead of Claude Haiku 4.5 at 18.2 (#105). GPT-5 mini leads on 7 of the 10 benchmarks both report. GPT-5 mini is 2.9x cheaper per token at list price. GPT-5 mini has the larger context window (400K tokens).
* Estimated: no published scores in that category. How the Respan Index works
| 31.8% |
| 43% |
| SimpleQA Verified | 13.2% | 21.6% |
| Vending-Bench 2 (Andon Labs) | 458.89 | -31.18 |
| WeirdML (Håvard Tveit Ihle) | 45.4% | 52.7% |
| LMArena Elo | 1414 | 1390 |
| SWE-bench Verified | 73.3% | 71% |
Higher is better. Each score links to where it was published.