Benchmarks · Science & research · Reasoning & puzzles · Vision
HLE-Diamond
Uses Artificial Analysis overall accuracy across all 1,000 questions, without tools. Score: accuracy. Collected by Jake Boggs' benchmark dashboard.
76 models · 145 results · data as of 2026-11-15 · artificialanalysis.ai ↗
Leaderboard
Fable 5.2 max
All results
145 rows| # | Model | Effort | HLE-Diamond |
|---|---|---|---|
| 1 | Fable 5.2 | max | 80.2% |
| 2 | Fable 5.2 | xhigh | 75.8% |
| 3 | Fable 5.2 | high | 73.8% |
| 4 | Gemini 4 Carbon | high | 69.1% |
| 5 | Opus 5.5 | max | 67.9% |
| 6 | Gemini 4 Carbon | medium | 67.4% |
| 7 | Fable 5.2 | medium | 66.3% |
| 8 | Opus 5.5 | xhigh | 64.0% |
| 9 | GPT-6 Astra | max | 63.8% |
| 10 | Gemini 4 Carbon | low | 61.9% |
| 11 | GPT-6 Astra | xhigh | 61.5% |
| 12 | Fable 5.2 | low | 60.8% |
| 13 | Fable 5.1 | max | 60.1% |
| 14 | GPT-6 Astra | high | 59.3% |
| 15 | Fable 5.1 | xhigh | 58.8% |
| 16 | Gemini 4 Hydrogen | xhigh | 58.8% |
| 17 | GPT-6.1 Sol | max | 58.8% |
| 18 | Gemini 4 Argon | high | 56.7% |
| 19 | GPT-6.1 Sol | xhigh | 55.8% |
| 20 | Opus 5.5 | high | 55.6% |
| 21 | Gemini 4 Hydrogen | high | 54.9% |
| 22 | Sonnet 5.5 | max | 53.3% |
| 23 | GPT-6 Astra | medium | 52.7% |
| 24 | GPT-6.1 Sol | high | 51.9% |
| 25 | Opus 5.5 | medium | 51.4% |
| 26 | Fable 5.1 | high | 50.4% |
| 27 | Muse Wonder Full | max | 48.5% |
| 28 | Gemini 4 Hydrogen | medium | 47.7% |
| 29 | GPT-6.1 Terra | max | 47.3% |
| 30 | GPT-6.1 Sol | medium | 46.0% |