Benchmarks · reasoning
ZeroBench
Uses main-question pass@5 because current ZeroBench evaluations no longer report pass@1. Score: main questions pass@5. Collected by Jake Boggs' benchmark dashboard.
61 models · 61 results · data as of 2026-10-29 · zerobench.github.io ↗
Leaderboard
GPT-6 Astra max
All results
61 rows| # | Model | Effort | ZeroBench |
|---|---|---|---|
| 1 | GPT-6 Astra | max | 52.0% |
| 2 | Opus 5.5 | max | 43.0% |
| 3 | GPT-6 Sol | max | 41.0% |
| 4 | GPT-5.6 Sol | max | 30.0% |
| 5 | Opus 5 | max | 26.0% |
| 6 | Fable 5 | max | 24.0% |
| 7 | GPT-5.4 | xhigh | 23.0% |
| 8 | GPT-5.5 | xhigh | 23.0% |
| 9 | Gemini 3.5 Flash | xhigh | 22.0% |
| 10 | GPT-5.6 Luna | max | 21.0% |
| 11 | Gemini 3 Pro | high | 20.0% |
| 12 | GPT-6 Luna | max | 20.0% |
| 13 | Gemini 3.1 Pro | high | 19.0% |
| 14 | GPT-5.6 Terra | max | 19.0% |
| 15 | Opus 4.8 | max | 18.0% |
| 16 | Gemini 3.6 Flash | high | 18.0% |
| 17 | GPT-5.2 | medium | 18.0% |
| 18 | Grok 4.6 | xhigh | 17.0% |
| 19 | Opus 4.7 | max | 15.0% |
| 20 | Gemini 3 Flash | high | 15.0% |
| 21 | Grok 4.5 | high | 15.0% |
| 22 | Gemini 3.5 Flash-Lite | high | 14.0% |
| 23 | Sonnet 5 | max | 13.0% |
| 24 | Grok 4.3 | high | 13.0% |
| 25 | Opus 4.6 | max | 12.0% |
| 26 | Sonnet 4.6 | max | 12.0% |
| 27 | GPT-5.4 Mini | xhigh | 11.0% |
| 28 | o4 mini | high | 11.0% |
| 29 | Grok 4.20 | xhigh | 10.0% |
| 30 | Opus 4.5 | high | 9.0% |