# Vals LegalBench (LegalBench)

> Vals AI legal benchmark with issue, rule, conclusion, interpretation, and rhetoric task views.

Canonical page: https://benchlm.ai/benchmarks/valslegalbench

- Category: [external](/external)
- Last updated: September 10, 2026

## About LegalBench

- Year: 2026
- Tasks: Legal reasoning task views
- Format: Accuracy score
- Difficulty: Professional legal reasoning
- Paper: [LegalBench](https://www.vals.ai/benchmarks/legal_bench)

BenchLM mirrors Vals LegalBench as display-only legal-domain evidence and does not use it in weighted rankings.

LegalBench is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (143 models)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Claude Fable 5](/models/claude-fable) | — | Anthropic | 88.56% |
| 2 | [Claude Fable 5.1](/models/claude-fable-5-1) | — | Anthropic | 88.51% |
| 3 | [Gemini 3.1 Pro Preview](https://www.vals.ai/models/google_gemini-3.1-pro-preview) | high reasoning | Google | 87.40% |
| 4 | [Gemini 3.7 Flash](/models/gemini-3-7-flash) | high reasoning | Google | 87.26% |
| 5 | [Gemini 3 Pro Preview](https://www.vals.ai/models/google_gemini-3-pro-preview) | high reasoning | Google | 87.03% |
| 6 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | high reasoning | Google | 86.99% |
| 7 | [Claude Opus 5](/models/claude-opus-5) | — | Anthropic | 86.97% |
| 8 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | max reasoning | OpenAI | 86.97% |
| 9 | [Gemini 3 Flash Preview](https://www.vals.ai/models/google_gemini-3-flash-preview) | high reasoning | Google | 86.86% |
| 10 | [Gemini 3.6 Flash](/models/gemini-3-6-flash) | high reasoning | Google | 86.70% |
| 11 | [GPT-5.5](/models/gpt-5-5) | xhigh reasoning | OpenAI | 86.52% |
| 12 | [Grok 4.6](/models/grok-4-6) | high reasoning | xAI | 86.31% |
| 13 | [GPT-5.4](/models/gpt-5-4) | xhigh reasoning | OpenAI | 86.04% |
| 14 | [GPT-5](https://www.vals.ai/models/openai_gpt-5-2025-08-07) | high reasoning | OpenAI | 86.02% |
| 15 | [Kimi K3](/models/kimi-k3) | — | Moonshot AI | 86.02% |
| 16 | [Grok 4.5](/models/grok-4-5) | high reasoning | xAI | 85.97% |
| 17 | [GPT-5.1](/models/gpt-5-1) | high reasoning | OpenAI | 85.68% |
| 18 | [MiniMax M3](/models/minimax-m3) | — | MiniMax | 85.42% |
| 19 | [Claude Opus 4.6 (Adaptive)](/models/claude-opus-4-6-thinking) | — | Anthropic | 85.30% |
| 20 | [Muse Spark 1.2](/models/muse-spark-1-2) | xhigh reasoning | Meta | 85.26% |
| 21 | [Claude Opus 4.7](/models/claude-opus-4-7) | — | Anthropic | 85.25% |
| 22 | [GPT-5.6 Terra](/models/gpt-5-6-terra) | max reasoning | OpenAI | 85.11% |
| 23 | [Qwen3.5 Plus Thinking](https://www.vals.ai/models/alibaba_qwen3.5-plus-thinking) | — | Alibaba | 85.10% |
| 24 | [Muse Spark 1.1](/models/muse-spark-1-1) | xhigh reasoning | Meta | 84.98% |
| 25 | [Qwen3.7 Max](/models/qwen3-7-max) | — | Alibaba | 84.91% |
| 26 | [GLM-5.3](/models/glm-5-3) | max reasoning | Z.AI | 84.84% |
| 27 | [Kimi K2.6](/models/kimi-2-6) | — | Moonshot AI | 84.74% |
| 28 | [Claude Opus 4.5 Thinking](/models/claude-opus-4-5-thinking) | — | Anthropic | 84.60% |
| 29 | [Grok 4.3](/models/grok-4-3) | — | xAI | 84.46% |
| 30 | [GLM-5.1](/models/glm-5-1) | — | Z.AI | 84.39% |
| 31 | [Gemini 2.5 Pro Exp 03 25](https://www.vals.ai/models/google_gemini-2.5-pro-exp-03-25) | — | Google | 84.32% |
| 32 | [Qwen3.5 Flash](/models/qwen3-5-flash) | — | Alibaba | 84.28% |
| 33 | [Qwen3.6 Plus](/models/qwen3-6-plus) | — | Alibaba | 84.23% |
| 34 | [Muse Spark](/models/muse-spark) | — | Meta | 84.22% |
| 35 | [Claude Sonnet 4.5 Thinking](/models/claude-sonnet-4-5-thinking) | — | Anthropic | 84.08% |
| 36 | [Gemini 3.5 Flash-Lite](/models/gemini-3-5-flash-lite) | high reasoning | Google | 84.07% |
| 37 | [GLM-5.2](/models/glm-5-2) | — | Z.AI | 84.07% |
| 38 | [GLM 5 Thinking](https://www.vals.ai/models/zai_glm-5-thinking) | — | Zhipu AI | 84.06% |
| 39 | [GPT-5.6 Luna](/models/gpt-5-6-luna) | max reasoning | OpenAI | 84.03% |
| 40 | [MiniMax M2.7](/models/minimax-m2-7) | — | MiniMax | 83.98% |
| 41 | [GLM-5.3-Flash](/models/glm-5-3-flash) | max reasoning | Z.AI | 83.93% |
| 42 | [Claude Sonnet 5](/models/claude-sonnet-5) | — | Anthropic | 83.92% |
| 43 | [Gemini 2.5 Flash Preview 04 17](https://www.vals.ai/models/google_gemini-2.5-flash-preview-04-17) | — | Google | 83.80% |
| 44 | [Gemini 3.1 Flash Lite Preview](https://www.vals.ai/models/google_gemini-3.1-flash-lite-preview) | high reasoning | Google | 83.76% |
| 45 | [o3](/models/o3) | high reasoning | OpenAI | 83.76% |
| 46 | [Qwen3.8 Max](/models/qwen3-8-max) | — | Alibaba | 83.61% |
| 47 | [Gemini 3.5 Flash](/models/gemini-3-5-flash) | high reasoning | Google | 83.60% |
| 48 | [Claude Opus 4.8](/models/claude-opus-4-8) | — | Anthropic | 83.57% |
| 49 | [Claude Opus 4.1](https://www.vals.ai/models/anthropic_claude-opus-4-1-20250805) | — | Anthropic | 83.46% |
| 50 | [GLM-4.7](/models/glm-4-7) | — | Z.AI | 83.36% |
| 51 | [DeepSeek V4.1 Flash](/models/deepseek-v4-1-flash) | high reasoning | DeepSeek | 83.28% |
| 52 | [Grok 4 0709](https://www.vals.ai/models/grok_grok-4-0709) | — | xAI | 83.19% |
| 53 | [Grok 3 Mini Fast High Reasoning](https://www.vals.ai/models/grok_grok-3-mini-fast-high-reasoning) | high reasoning | xAI | 83.14% |
| 54 | [GPT-4.1](/models/gpt-4-1) | high reasoning | OpenAI | 83.10% |
| 55 | [Claude Opus 4](https://www.vals.ai/models/anthropic_claude-opus-4-20250514) | — | Anthropic | 83.07% |
| 56 | [Inkling-Small](/models/inkling-small) | 0.99 reasoning | Thinking Machines Lab | 82.99% |
| 57 | [Claude Sonnet 4](https://www.vals.ai/models/anthropic_claude-sonnet-4-20250514) | — | Anthropic | 82.95% |
| 58 | [Inkling](/models/inkling) | 0.99 reasoning | Thinking Machines Lab | 82.95% |
| 59 | [Claude Opus 4.5](/models/claude-opus-4-5) | — | Anthropic | 82.84% |
| 60 | [GPT-5.2](/models/gpt-5-2) | xhigh reasoning | OpenAI | 82.76% |
| 61 | [Gemini 2.5 Flash Preview 09 2025 Thinking](https://www.vals.ai/models/google_gemini-2.5-flash-preview-09-2025-thinking) | — | Google | 82.63% |
| 62 | [Grok 3](https://www.vals.ai/models/grok_grok-3) | — | xAI | 82.59% |
| 63 | [Claude 3.7 Sonnet 20250219 Thinking](https://www.vals.ai/models/anthropic_claude-3-7-sonnet-20250219-thinking) | — | Anthropic | 82.52% |
| 64 | [Gemini 2.5 Flash Preview 09 2025](https://www.vals.ai/models/google_gemini-2.5-flash-preview-09-2025) | — | Google | 82.49% |
| 65 | [Grok 4.1 Fast (Reasoning)](/models/grok-4-1-fast-reasoning) | — | xAI | 82.45% |
| 66 | [Qwen3.8-27B](/models/qwen3-8-27b) | xhigh reasoning | Alibaba | 82.43% |
| 67 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | max reasoning | DeepSeek | 82.36% |
| 68 | [GPT-4o](/models/gpt-4o) | high reasoning | OpenAI | 82.21% |
| 69 | [Claude Sonnet 4.6](/models/claude-sonnet-4-6) | — | Anthropic | 82.12% |
| 70 | [Nemotron 3 Ultra 550b A55b](https://www.vals.ai/models/nvidia_nemotron-3-ultra-550b-a55b) | — | Nvidia | 82.07% |
| 71 | [Claude Sonnet 4 20250514 Thinking](https://www.vals.ai/models/anthropic_claude-sonnet-4-20250514-thinking) | — | Anthropic | 82.06% |
| 72 | [Gemini 2.5 Flash Lite Preview 09 2025 Thinking](https://www.vals.ai/models/google_gemini-2.5-flash-lite-preview-09-2025-thinking) | — | Google | 82.04% |
| 73 | [Grok 3 Mini Fast Low Reasoning](https://www.vals.ai/models/grok_grok-3-mini-fast-low-reasoning) | low reasoning | xAI | 81.92% |
| 74 | [Qwen3 Max](/models/qwen3-max) | — | Alibaba | 81.86% |
| 75 | [GPT-5 mini](/models/gpt-5-mini) | high reasoning | OpenAI | 81.77% |
| 76 | [Moonshotai Kimi K2 Instruct](https://www.vals.ai/models/together_moonshotai/Kimi-K2-Instruct) | — | Together AI | 81.45% |
| 77 | [Claude Haiku 4.5 Thinking](/models/claude-haiku-4-5-thinking) | — | Anthropic | 81.24% |
| 78 | [DeepSeek V3](/models/deepseek-v3) | — | DeepSeek | 80.76% |
| 79 | [Grok 4 Fast (Reasoning)](/models/grok-4-fast-reasoning) | — | xAI | 80.60% |
| 80 | [GPT-4 Turbo](/models/gpt-4-turbo) | high reasoning | OpenAI | 80.46% |
| 81 | [o1](/models/o1) | high reasoning | OpenAI | 80.39% |
| 82 | [Qwen3 Max Preview](https://www.vals.ai/models/alibaba_qwen3-max-preview) | — | Alibaba | 80.33% |
| 83 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | max reasoning | DeepSeek | 80.32% |
| 84 | [Kimi K2 Thinking](https://www.vals.ai/models/kimi_kimi-k2-thinking) | — | Moonshot AI | 80.20% |
| 85 | [Qwen3 235b A22b](https://www.vals.ai/models/fireworks_qwen3-235b-a22b) | — | Fireworks AI | 80.18% |
| 86 | [GPT-4o](/models/gpt-4o) | high reasoning | OpenAI | 80.12% |
| 87 | [Claude 3.7 Sonnet](https://www.vals.ai/models/anthropic_claude-3-7-sonnet-20250219) | — | Anthropic | 80.00% |
| 88 | [MiniMax M2.5](/models/minimax-m2-5) | — | MiniMax | 79.96% |
| 89 | [Command A 03 2025](https://www.vals.ai/models/cohere_command-a-03-2025) | — | Cohere | 79.70% |
| 90 | [Ling 3.0 Flash 2607](https://www.vals.ai/models/ant_ling-3.0-flash-2607) | — | Ant | 79.69% |
| 91 | [GLM-4.6](/models/glm-4-6) | — | Z.AI | 79.61% |
| 92 | [Qwen Qwen2.5 72B Instruct Turbo](https://www.vals.ai/models/together_Qwen/Qwen2.5-72B-Instruct-Turbo) | — | Together AI | 79.40% |
| 93 | [O4 Mini](https://www.vals.ai/models/openai_o4-mini-2025-04-16) | high reasoning | OpenAI | 79.19% |
| 94 | [Mistral Large 2512](https://www.vals.ai/models/mistralai_mistral-large-2512) | — | Mistral AI | 79.14% |
| 95 | [Grok 4.1 Fast Non Reasoning](https://www.vals.ai/models/grok_grok-4-1-fast-non-reasoning) | — | xAI | 79.11% |
| 96 | [Gemini 2.5 Flash Lite Preview 09 2025](https://www.vals.ai/models/google_gemini-2.5-flash-lite-preview-09-2025) | — | Google | 79.01% |
| 97 | [MiMo-V2.5](/models/mimo-v2-5) | — | Xiaomi | 78.89% |
| 98 | [Grok 4 Fast Non Reasoning](https://www.vals.ai/models/grok_grok-4-fast-non-reasoning) | — | xAI | 78.40% |
| 99 | [Gemini 2.0 Flash 001](https://www.vals.ai/models/google_gemini-2.0-flash-001) | — | Google | 78.36% |
| 100 | [GPT-4.1 mini](/models/gpt-4-1-mini) | high reasoning | OpenAI | 78.04% |
| 101 | [GPT-5.4 nano](/models/gpt-5-4-nano) | high reasoning | OpenAI | 77.92% |
| 102 | [Llama4 Maverick Instruct Basic](https://www.vals.ai/models/fireworks_llama4-maverick-instruct-basic) | — | Fireworks AI | 77.81% |
| 103 | [Grok 4.20 0309 Reasoning](https://www.vals.ai/models/grok_grok-4.20-0309-reasoning) | — | xAI | 77.74% |
| 104 | [DeepSeek V3 0324](https://www.vals.ai/models/fireworks_deepseek-v3-0324) | — | Fireworks AI | 77.73% |
| 105 | [DeepSeek V4 Flash 0731](/models/deepseek-v4-flash-0731) | high reasoning | DeepSeek | 77.71% |
| 106 | [Meta Llama Llama 3.3 70B Instruct Turbo](https://www.vals.ai/models/together_meta-llama/Llama-3.3-70B-Instruct-Turbo) | — | Together AI | 77.18% |
| 107 | [MiMo-V2.5-Pro](/models/mimo-v2-5-pro) | — | Xiaomi | 77.13% |
| 108 | [DeepSeek V3p2 Thinking](https://www.vals.ai/models/fireworks_deepseek-v3p2-thinking) | — | Fireworks AI | 76.08% |
| 109 | [GPT-OSS 120B](/models/gpt-oss-120b) | — | OpenAI | 75.94% |
| 110 | [GLM-4.5](/models/glm-4-5) | — | Z.AI | 75.63% |
| 111 | [MiniMax M2.1](https://www.vals.ai/models/minimax_MiniMax-M2.1) | — | MiniMax | 75.45% |
| 112 | [Laguna M.1](/models/laguna-m-1) | — | Poolside | 75.14% |
| 113 | [Jamba 1.5 Large](https://www.vals.ai/models/ai21labs_jamba-1.5-large) | — | AI21 Labs | 74.16% |
| 114 | [Meta Llama Llama 4 Scout 17B 16E Instruct](https://www.vals.ai/models/together_meta-llama/Llama-4-Scout-17B-16E-Instruct) | — | Together AI | 72.04% |
| 115 | [Jamba Large 1.6](https://www.vals.ai/models/ai21labs_jamba-large-1.6) | — | AI21 Labs | 71.96% |
| 116 | [o3-mini](/models/o3-mini) | high reasoning | OpenAI | 71.54% |
| 117 | [Laguna XS.2](/models/laguna-xs-2) | — | Poolside | 71.03% |
| 118 | [GPT-OSS 20B](/models/gpt-oss-20b) | — | OpenAI | 70.85% |
| 119 | [Claude 3.5 Haiku](https://www.vals.ai/models/anthropic_claude-3-5-haiku-20241022) | — | Anthropic | 70.33% |
| 120 | [Gemini 1.0 Pro 002](https://www.vals.ai/models/google_gemini-1.0-pro-002) | — | Google | 70.25% |
| 121 | [Jamba Mini 1.6](https://www.vals.ai/models/ai21labs_jamba-mini-1.6) | — | AI21 Labs | 69.73% |
| 122 | [Qwen Qwen2.5 7B Instruct Turbo](https://www.vals.ai/models/together_Qwen/Qwen2.5-7B-Instruct-Turbo) | — | Together AI | 69.56% |
| 123 | [Mistral Small 2503](https://www.vals.ai/models/mistralai_mistral-small-2503) | — | Mistral AI | 69.16% |
| 124 | [Gemini 1.5 Pro 002](https://www.vals.ai/models/google_gemini-1.5-pro-002) | — | Google | 69.08% |
| 125 | [Command R Plus](https://www.vals.ai/models/cohere_command-r-plus) | — | Cohere | 68.29% |
| 126 | [Google Gemma 2 9b It](https://www.vals.ai/models/together_google/gemma-2-9b-it) | — | Together AI | 68.14% |
| 127 | [Google Gemma 2 27b It](https://www.vals.ai/models/together_google/gemma-2-27b-it) | — | Together AI | 67.98% |
| 128 | [DeepSeek-R1](/models/deepseek-r1) | — | DeepSeek | 67.32% |
| 129 | [Jamba 1.5 Mini](https://www.vals.ai/models/ai21labs_jamba-1.5-mini) | — | AI21 Labs | 66.62% |
| 130 | [DeepSeek V3p2](https://www.vals.ai/models/fireworks_deepseek-v3p2) | — | Fireworks AI | 65.96% |
| 131 | [GPT-3.5 Turbo](https://www.vals.ai/models/openai_gpt-3.5-turbo) | high reasoning | OpenAI | 64.37% |
| 132 | [Gemini 1.5 Flash 002](https://www.vals.ai/models/google_gemini-1.5-flash-002) | — | Google | 63.24% |
| 133 | [Meta Llama Llama 2 70b Hf](https://www.vals.ai/models/together_meta-llama/Llama-2-70b-hf) | — | Together AI | 62.43% |
| 134 | [Mistral Medium 2505](https://www.vals.ai/models/mistralai_mistral-medium-2505) | — | Mistral AI | 61.98% |
| 135 | [GPT-4.1 nano](/models/gpt-4-1-nano) | high reasoning | OpenAI | 61.06% |
| 136 | [Mistralai Mixtral 8x7B V0.1](https://www.vals.ai/models/together_mistralai/Mixtral-8x7B-v0.1) | — | Together AI | 55.84% |
| 137 | [Magistral Medium 2509](https://www.vals.ai/models/mistralai_magistral-medium-2509) | — | Mistral AI | 54.77% |
| 138 | [Mistralai Mistral 7B V0.1](https://www.vals.ai/models/together_mistralai/Mistral-7B-v0.1) | — | Together AI | 53.77% |
| 139 | [Togethercomputer Llama 2 13b](https://www.vals.ai/models/together_togethercomputer/llama-2-13b) | — | Together AI | 53.70% |
| 140 | [Togethercomputer Llama 2 7b](https://www.vals.ai/models/together_togethercomputer/llama-2-7b) | — | Together AI | 51.87% |
| 141 | [GPT-5 nano](/models/gpt-5-nano) | high reasoning | OpenAI | 50.13% |
| 142 | [Magistral Small 2509](https://www.vals.ai/models/mistralai_magistral-small-2509) | — | Mistral AI | 40.02% |
| 143 | [Command R](https://www.vals.ai/models/cohere_command-r) | — | Cohere | 32.97% |

## FAQ

### What does LegalBench measure?

Vals AI legal benchmark with issue, rule, conclusion, interpretation, and rhetoric task views.

### Which model leads the published LegalBench snapshot?

Claude Fable 5 currently leads the published LegalBench snapshot with a score of 88.56%.

### How many models are evaluated on LegalBench?

The September 10, 2026 contains 143 AI models.

### Does LegalBench affect BenchLM's overall score?

Not directly. LegalBench is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.