Benchmarks · reasoning
ARC-AGI-1
ARC-AGI-1 semi-private evaluation set: novel grid-transformation puzzles measuring fluid intelligence; percent of tasks solved, with verified cost per task.
84 models · 232 results · data as of 2026-10-29 · arcprize.org ↗
Leaderboard
Fable 5 max
All results
232 rows| # | Model | Effort | ARC-AGI-1 | Subscription $ / task | API $ / task |
|---|---|---|---|---|---|
| 1 | Fable 5 | max | 98.5% | $0.24 | $2.11 |
| 2 | Fable 5 | xhigh | 98.5% | $0.12 | $1.02 |
| 3 | Opus 5.5 | high | 98.5% | $0.002 | $0.16 |
| 4 | Gemini 3.8 Flash | high | 98.5% | $0.008 | $0.21 |
| 5 | Gemini 4 Argon | high | 98.5% | $0.058 | $0.16 |
| 6 | Gemini 4 Argon | medium | 98.5% | $0.023 | $0.065 |
| 7 | GPT-6 Astra | high | 98.5% | $0.024 | $0.28 |
| 8 | GPT-6 Astra | xhigh | 98.5% | $0.030 | $0.35 |
| 9 | GPT-6.1 Sol | high | 98.5% | $0.007 | $0.059 |
| 10 | GPT-6.1 Sol | xhigh | 98.5% | $0.009 | $0.074 |
| 11 | Gemini 3.1 Pro | high | 98.0% | $0.52 | $0.52 |
| 12 | Fable 5.1 | max | 97.5% | $0.22 | $1.40 |
| 13 | Opus 5.5 | max | 97.5% | $0.010 | $0.64 |
| 14 | Opus 5.5 | medium | 97.5% | $0.002 | $0.15 |
| 15 | Opus 5.5 | xhigh | 97.5% | $0.003 | $0.22 |
| 16 | Opus 5 | high | 97.5% | $0.008 | $0.47 |
| 17 | Opus 5 | max | 97.5% | $0.012 | $0.70 |
| 18 | Gemini 3.8 Flash | medium | 97.5% | $0.004 | $0.11 |
| 19 | GPT-5.6 Sol | xhigh | 97.5% | $0.022 | $0.40 |
| 20 | GPT-6 Astra | max | 97.5% | $0.037 | $0.43 |
| 21 | GPT-6 Astra | medium | 97.5% | $0.021 | $0.24 |
| 22 | GPT-5.6 Sol | high | 97.0% | $0.017 | $0.30 |
| 23 | Fable 5.1 | xhigh | 96.5% | $0.15 | $0.96 |
| 24 | Sonnet 5.5 | max | 96.5% | $0.008 | $0.50 |
| 25 | Sonnet 5.5 | xhigh | 96.5% | $0.002 | $0.14 |
| 26 | GPT-5.5 Pro | high | 96.5% | $4.53 | $4.53 |
| 27 | GPT-5.6 Sol | max | 96.5% | $0.030 | $0.54 |
| 28 | GPT-5.6 Terra | max | 96.5% | $0.030 | $0.55 |
| 29 | GPT-6 Astra | low | 96.5% | $0.019 | $0.23 |
| 30 | GPT-6.1 Sol | max | 96.5% | $0.011 | $0.095 |