Share promptsSign inExplore benchmarks
← Legal benchmarks

LegalBench leaderboard (Oct 2026): AI model scores

As of 13 Oct 2026, the highest reported OpenAI score on LegalBench is GPT-5.6 Sol (OpenAI) at 86.97%, #10 on that leaderboard, per Vals AI.

What it measures

A wide range of open-source legal reasoning tasks.

Open-source legal reasoning tasks such as issue spotting, rule recall and contract interpretation, scored on accuracy.

What the score means

Accuracy

Share of LegalBench questions answered correctly.

Results from published AI leaderboards. Compare models within each benchmark; scores from different tests aren’t directly comparable. How we collect these results

Scores by model

Run by Vals AI. Leaderboard dated 13 Oct 2026. Last checked 22 Oct 2026. Showing top 25 of 151 models listed.

Highlight a provider: Anthropic OpenAI Google xAI Meta Z.ai Moonshot MiniMax

Highlighting OpenAI models. Other providers are dimmed. Show all providers

  • Anthropic
  • OpenAI
  • Google
  • xAI
  • Meta
  • Z.ai
  • Moonshot
  • MiniMax
LegalBench: top 15 by AccuracyBars from zero. Leader: Claude Fable 5 at 88.56%. OpenAI models highlighted. Full values in the table below.0%20%40%60%80%100%Claude Fable 5 (Anthropic), rank 1: 88.56%AClaude Fable 588.56%Claude Fable 5.1 (Anthropic), rank 2: 88.51%AClaude Fable 5.188.51%Gemini 4 Argon (Google), rank 3: 88.30%GGemini 4 Argon88.30%Wonder Full (Meta), rank 4: 87.42%MWonder Full87.42%Gemini 3.1 Pro Preview (02/26) (Google), rank 5: 87.40%GGemini 3.1 Pro Preview87.40%Gemini 3.7 Flash (Google), rank 6: 87.26%GGemini 3.7 Flash87.26%Gemini 3 Pro (11/25) (Google), rank 7: 87.03%GGemini 3 Pro87.03%Gemini 3.8 Flash (Google), rank 8: 86.99%GGemini 3.8 Flash86.99%Claude Opus 5 (Anthropic), rank 9: 86.97%AClaude Opus 586.97%GPT-5.6 Sol (OpenAI), rank 10: 86.97%OGPT-5.6 Sol86.97%Gemini 3 Flash (12/25) (Google), rank 11: 86.86%GGemini 3 Flash86.86%Gemini 3.6 Flash (Google), rank 12: 86.70%GGemini 3.6 Flash86.70%GPT 5.5 (OpenAI), rank 13: 86.52%OGPT 5.586.52%Grok 4.6 (xAI), rank 14: 86.31%XGrok 4.686.31%GLM Legal (Z.ai), rank 15: 86.23%ZGLM Legal86.23%GPT 5.4 (xhigh) (OpenAI), rank 17: 86.04%O#17 GPT 5.486.04%GPT 5 (OpenAI), rank 18: 86.02%O#18 GPT 586.02%GPT 5.1 (OpenAI), rank 20: 85.68%O#20 GPT 5.185.68%
LegalBench: top 12 by AccuracyBars from zero. Leader: Claude Fable 5 at 88.56%. OpenAI models highlighted. Full values in the table below.0%50%100%Claude Fable 5 (Anthropic), rank 1: 88.56%AClaude Fable 588.56%Claude Fable 5.1 (Anthropic), rank 2: 88.51%AClaude Fable 5.188.51%Gemini 4 Argon (Google), rank 3: 88.30%GGemini 4 Argon88.30%Wonder Full (Meta), rank 4: 87.42%MWonder Full87.42%Gemini 3.1 Pro Preview (02/26) (Google), rank 5: 87.40%GGemini 3.1 Pro P…87.40%Gemini 3.7 Flash (Google), rank 6: 87.26%GGemini 3.7 Flash87.26%Gemini 3 Pro (11/25) (Google), rank 7: 87.03%GGemini 3 Pro87.03%Gemini 3.8 Flash (Google), rank 8: 86.99%GGemini 3.8 Flash86.99%Claude Opus 5 (Anthropic), rank 9: 86.97%AClaude Opus 586.97%GPT-5.6 Sol (OpenAI), rank 10: 86.97%OGPT-5.6 Sol86.97%Gemini 3 Flash (12/25) (Google), rank 11: 86.86%GGemini 3 Flash86.86%Gemini 3.6 Flash (Google), rank 12: 86.70%GGemini 3.6 Flash86.70%GPT 5.5 (OpenAI), rank 13: 86.52%O#13 GPT 5.586.52%GPT 5.4 (xhigh) (OpenAI), rank 17: 86.04%O#17 GPT 5.486.04%GPT 5 (OpenAI), rank 18: 86.02%O#18 GPT 586.02%
LegalBench: top 15 by AccuracyDot plot, axis starting at 80% (not zero). Leader: Claude Fable 5 at 88.56%. OpenAI models highlighted. Full values in the table below.80%82%84%86%88%90%ACCURACY · AXIS STARTS AT 80%, NOT ZEROClaude Fable 5 (Anthropic), rank 1: 88.56%AClaude Fable 588.56%Claude Fable 5.1 (Anthropic), rank 2: 88.51%AClaude Fable 5.188.51%Gemini 4 Argon (Google), rank 3: 88.30%GGemini 4 Argon88.30%Wonder Full (Meta), rank 4: 87.42%MWonder Full87.42%Gemini 3.1 Pro Preview (02/26) (Google), rank 5: 87.40%GGemini 3.1 Pro Preview87.40%Gemini 3.7 Flash (Google), rank 6: 87.26%GGemini 3.7 Flash87.26%Gemini 3 Pro (11/25) (Google), rank 7: 87.03%GGemini 3 Pro87.03%Gemini 3.8 Flash (Google), rank 8: 86.99%GGemini 3.8 Flash86.99%Claude Opus 5 (Anthropic), rank 9: 86.97%AClaude Opus 586.97%GPT-5.6 Sol (OpenAI), rank 10: 86.97%OGPT-5.6 Sol86.97%Gemini 3 Flash (12/25) (Google), rank 11: 86.86%GGemini 3 Flash86.86%Gemini 3.6 Flash (Google), rank 12: 86.70%GGemini 3.6 Flash86.70%GPT 5.5 (OpenAI), rank 13: 86.52%OGPT 5.586.52%Grok 4.6 (xAI), rank 14: 86.31%XGrok 4.686.31%GLM Legal (Z.ai), rank 15: 86.23%ZGLM Legal86.23%GPT 5.4 (xhigh) (OpenAI), rank 17: 86.04%O#17 GPT 5.486.04%GPT 5 (OpenAI), rank 18: 86.02%O#18 GPT 586.02%GPT 5.1 (OpenAI), rank 20: 85.68%O#20 GPT 5.185.68%
LegalBench: top 12 by AccuracyDot plot, axis starting at 80% (not zero). Leader: Claude Fable 5 at 88.56%. OpenAI models highlighted. Full values in the table below.80%85%90%AXIS STARTS AT 80%, NOT ZEROClaude Fable 5 (Anthropic), rank 1: 88.56%AClaude Fable 588.56%Claude Fable 5.1 (Anthropic), rank 2: 88.51%AClaude Fable 5.188.51%Gemini 4 Argon (Google), rank 3: 88.30%GGemini 4 Argon88.30%Wonder Full (Meta), rank 4: 87.42%MWonder Full87.42%Gemini 3.1 Pro Preview (02/26) (Google), rank 5: 87.40%GGemini 3.1 Pro P…87.40%Gemini 3.7 Flash (Google), rank 6: 87.26%GGemini 3.7 Flash87.26%Gemini 3 Pro (11/25) (Google), rank 7: 87.03%GGemini 3 Pro87.03%Gemini 3.8 Flash (Google), rank 8: 86.99%GGemini 3.8 Flash86.99%Claude Opus 5 (Anthropic), rank 9: 86.97%AClaude Opus 586.97%GPT-5.6 Sol (OpenAI), rank 10: 86.97%OGPT-5.6 Sol86.97%Gemini 3 Flash (12/25) (Google), rank 11: 86.86%GGemini 3 Flash86.86%Gemini 3.6 Flash (Google), rank 12: 86.70%GGemini 3.6 Flash86.70%GPT 5.5 (OpenAI), rank 13: 86.52%O#13 GPT 5.586.52%GPT 5.4 (xhigh) (OpenAI), rank 17: 86.04%O#17 GPT 5.486.04%GPT 5 (OpenAI), rank 18: 86.02%O#18 GPT 586.02%
Accuracy
Share of LegalBench questions answered correctly.

All scores

LegalBench scores, OpenAI models highlighted
RankModelProviderAccuracyPrice in / out ($/1M tokens)
1 Claude Fable 5Anthropic88.56%$10 / $50
2 Claude Fable 5.1Anthropic88.51%$10 / $50
3 Gemini 4 ArgonGoogle88.30%$4 / $20
4 Wonder FullMeta87.42%$4 / $20
5 Gemini 3.1 Pro Preview (02/26)Google87.40%$2 / $12
6 Gemini 3.7 FlashGoogle87.26%$1.5 / $7.5
7 Gemini 3 Pro (11/25)Google87.03%$2 / $12
8 Gemini 3.8 FlashGoogle86.99%$1.5 / $7.5
9 Claude Opus 5Anthropic86.97%$5 / $25
10 GPT-5.6 SolOpenAI86.97%$4 / $20
Show 15 more rows
RankModelProviderAccuracyPrice in / out ($/1M tokens)
11 Gemini 3 Flash (12/25)Google86.86%$0.5 / $3
12 Gemini 3.6 FlashGoogle86.70%$1.5 / $7.5
13 GPT 5.5OpenAI86.52%$5 / $30
14 Grok 4.6xAI86.31%$2 / $6
15 GLM LegalZ.ai86.23%$0.075 / $0.25
16 Kimi K3Moonshot86.21%$3 / $15
17 GPT 5.4 (xhigh)OpenAI86.04%$2.5 / $15
18 GPT 5OpenAI86.02%$1.25 / $10
19 Grok 4.5xAI85.97%$2 / $6
20 GPT 5.1OpenAI85.68%$1.25 / $10
21 MiniMax-M3MiniMax85.42%$0.6 / $2.4
22 Claude Opus 4.6 (Thinking)Anthropic85.30%$5 / $25
23 Muse Spark 1.2Meta85.26%$1.25 / $4.25
24 Claude Opus 4.7Anthropic85.25%$5 / $25
25 GPT-5.6 TerraOpenAI85.11%$2 / $12

Source: Vals AI leaderboard ↗

Best-ranked OpenAI model here: GPT-5.6 Sol (OpenAI). Try prompts for ChatGPT →