Benchmarks · reasoning
LegalBench
Evaluating language models on a wide range of open source legal reasoning tasks.
63 models · 63 results · updated 2026-10-13 · vals.ai ↗
Leaderboard
Fable 5 max
More from this benchmark
1 published measuresTime per task150 results · lower is better
All results
63 rows| # | Model | Effort | LegalBench | Subscription $ / task | API $ / task | Time |
|---|---|---|---|---|---|---|
| 1 | Fable 5 | max | 88.6% | $0.003 | $0.031 | 9.0s |
| 2 | Fable 5.1 | max | 88.5% | $0.008 | $0.052 | 12s |
| 3 | Gemini 4 Argon | high | 88.3% | $0.006 | $0.016 | 13s |
| 4 | Gemini 3.7 Flash | high | 87.3% | $0.001 | $0.004 | 2.2s |
| 5 | Gemini 3.8 Flash | high | 87.0% | $0.000 | $0.008 | 3.3s |
| 6 | Opus 5 | max | 87.0% | $0.000 | $0.012 | 4.4s |
| 7 | GPT-5.6 Sol | max | 87.0% | $0.001 | $0.013 | 6.2s |
| 8 | Gemini 3.6 Flash | high | 86.7% | $0.000 | $0.005 | 3.0s |
| 9 | GPT-5.5 | xhigh | 86.5% | $0.001 | $0.015 | 18s |
| 10 | Grok 4.6 | high | 86.3% | $0.001 | $0.015 | 27s |
| 11 | Kimi K3 | max | 86.2% | $0.001 | $0.008 | 7.1s |
| 12 | Grok 4.5 | high | 86.0% | $0.001 | $0.005 | 68s |
| 13 | MiniMax M3 | max | 85.4% | $0.000 | $0.001 | 8.2s |
| 14 | Muse Spark 1.2 | xhigh | 85.3% | $0.003 | $0.006 | 30s |
| 15 | Opus 4.7 | max | 85.3% | $0.008 | $0.008 | 27s |
| 16 | GPT-5.6 Terra | max | 85.1% | $0.000 | $0.003 | 2.8s |
| 17 | Muse Spark 1.1 | xhigh | 85.0% | $0.009 | $0.009 | 17s |
| 18 | Qwen 3.7 Max | max | 84.9% | $0.002 | $0.006 | 12s |
| 19 | GLM 5.3 | max | 84.8% | $0.001 | $0.007 | 22s |
| 20 | Kimi K2.6 | max | 84.7% | $0.006 | $0.006 | 50s |
| 21 | Grok 4.3 | xhigh | 84.5% | $0.003 | $0.003 | 5.8 min |
| 22 | GLM 5.1 | max | 84.4% | $0.002 | $0.002 | 14s |
| 23 | Grok 4.7 | xhigh | 84.4% | $0.001 | $0.011 | 26s |
| 24 | Qwen 3.5 Flash | max | 84.3% | $0.001 | $0.001 | 17s |
| 25 | Qwen 3.6 Plus | max | 84.2% | $0.003 | $0.003 | 19s |
| 26 | Muse Spark | 84.2% | $0.001 | $0.001 | 34s | |
| 27 | Gemini 3.5 Flash-Lite | high | 84.1% | $0.001 | $0.001 | 1.7s |
| 28 | GLM 5.2 | max | 84.1% | $0.000 | $0.002 | 5.8s |
| 29 | GPT-5.6 Luna | max | 84.0% | $0.000 | $0.001 | 5.9s |
| 30 | MiniMax M2.7 | high | 84.0% | $0.000 | $0.001 | 8.9s |