# Vals Harvey's Legal Agent Benchmark (Harvey's Legal Agent Benchmark)

> Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools

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

- Category: [Agentic](/agentic)
- Last updated: October 12, 2026

Evaluation results published by [Vals AI](https://www.vals.ai/benchmarks/hlab). Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.

## About Harvey's Legal Agent Benchmark

- Year: 2026
- Tasks: Legal agent work across documents, spreadsheets, presentations, and files
- Format: Accuracy score
- Difficulty: Professional legal workflow automation
- Paper: [Harvey's Legal Agent Benchmark](https://www.vals.ai/benchmarks/hlab)

BenchLM mirrors the public Vals AI Harvey's Legal Agent Benchmark leaderboard as display-only external evidence. The captured snapshot preserves overall scores, task-level scores where Vals publishes them, uncertainty, latency, and cost-per-test metadata. It is excluded from BenchLM weighted rankings.

Harvey's Legal Agent Benchmark is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (79 models)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Muse Spark 1.2](/models/muse-spark-1-2) | xhigh reasoning | Meta | 25.42% |
| 2 | [Muse Spark 1.3 Max](https://www.vals.ai/models/meta_muse_spark_1_3_max) | max reasoning | Meta | 23.75% |
| 3 | [Muse Spark 1.3](/models/muse-spark-1-3) | xhigh reasoning | Meta | 22.92% |
| 4 | [Muse Wonder Full](/models/muse-wonder-full) | max reasoning | Meta | 21.67% |
| 5 | [Muse Spark 1.1](/models/muse-spark-1-1) | xhigh reasoning | Meta | 20.00% |
| 6 | [Gemini 4 Argon](/models/gemini-4-argon) | high reasoning | Google | 19.58% |
| 7 | [GLM Legal](https://www.vals.ai/models/zai_glm-legal) | max reasoning | Zhipu AI | 19.17% |
| 8 | [Muse Wonder](/models/muse-wonder) | max reasoning | Meta | 17.08% |
| 9 | [Mistral Large 4](/models/mistral-large-4) | high reasoning | Mistral | 15.83% |
| 10 | [Grok 4.6](/models/grok-4-6) | high reasoning | xAI | 15.83% |
| 11 | [Grok 4.5](/models/grok-4-5) | high reasoning | xAI | 12.92% |
| 12 | [Kimi K3](/models/kimi-k3) | max reasoning | Moonshot AI | 12.92% |
| 13 | [Grok 4.7](/models/grok-4-7) | xhigh reasoning | xAI | 12.50% |
| 14 | [MiMo-V2.6-Flash](/models/mimo-v2-6-flash) | — | Xiaomi | 11.25% |
| 15 | [Qwen3.8-27B](/models/qwen3-8-27b) | xhigh reasoning | Alibaba | 11.25% |
| 16 | [Claude Fable 5](/models/claude-fable) | — | Anthropic | 11.25% |
| 17 | [MiMo-V2.6-Pro](/models/mimo-v2-6-pro) | — | Xiaomi | 10.83% |
| 18 | [Qwen3.8 Max](/models/qwen3-8-max) | — | Alibaba | 10.42% |
| 19 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | high reasoning | Google | 10.00% |
| 20 | [Claude Opus 4.8](/models/claude-opus-4-8) | — | Anthropic | 9.58% |
| 21 | [Step 5 Preview](/models/step-5-preview) | — | StepFun | 9.17% |
| 22 | [Hy4 preview](/models/hy4-preview) | — | Tencent | 9.17% |
| 23 | [Ember-1](/models/ember-1) | — | Fireworks | 9.17% |
| 24 | [Gemini 3.7 Flash](/models/gemini-3-7-flash) | high reasoning | Google | 8.75% |
| 25 | [DeepSeek V4 Flash 0731](/models/deepseek-v4-flash-0731) | high reasoning | DeepSeek | 8.33% |
| 26 | [GLM-5.3](/models/glm-5-3) | max reasoning | Z.AI | 8.33% |
| 27 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | max reasoning | DeepSeek | 7.50% |
| 28 | [GLM-5.2](/models/glm-5-2) | max reasoning | Z.AI | 7.08% |
| 29 | [DeepSeek V4.1 Flash 1017](/models/deepseek-v4-1-flash) | max reasoning | DeepSeek | 6.67% |
| 30 | [Claude Opus 4.7](/models/claude-opus-4-7) | — | Anthropic | 6.67% |
| 31 | [GLM-5.3-Flash](/models/glm-5-3-flash) | max reasoning | Z.AI | 6.67% |
| 32 | [Claude Opus 5](/models/claude-opus-5) | — | Anthropic | 6.67% |
| 33 | [Claude Fable 5.1](/models/claude-fable-5-1) | — | Anthropic | 6.67% |
| 34 | [GPT-6 Astra](/models/gpt-6-astra) | max reasoning | OpenAI | 5.42% |
| 35 | [GPT-6.1 Sol](/models/gpt-6-1-sol) | max reasoning | OpenAI | 5.42% |
| 36 | [Claude Sonnet 4.6](/models/claude-sonnet-4-6) | — | Anthropic | 5.00% |
| 37 | [Claude Sonnet 5](/models/claude-sonnet-5) | — | Anthropic | 5.00% |
| 38 | [MiniMax M3](/models/minimax-m3) | — | MiniMax | 4.17% |
| 39 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | max reasoning | DeepSeek | 3.75% |
| 40 | [GPT-5.5](/models/gpt-5-5) | xhigh reasoning | OpenAI | 3.75% |
| 41 | [Claude Opus 5.5](/models/claude-opus-5-5) | — | Anthropic | 3.75% |
| 42 | [Gemini 3.6 Flash](/models/gemini-3-6-flash) | high reasoning | Google | 3.33% |
| 43 | [GPT-6 Luna](/models/gpt-6-luna) | max reasoning | OpenAI | 2.92% |
| 44 | [Claude Sonnet 5.5](/models/claude-sonnet-5-5) | — | Anthropic | 2.92% |
| 45 | [Gemini 3.5 Flash](/models/gemini-3-5-flash) | high reasoning | Google | 2.50% |
| 46 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | max reasoning | OpenAI | 2.50% |
| 47 | [MiMo-V2.5-Pro](/models/mimo-v2-5-pro) | — | Xiaomi | 2.08% |
| 48 | [Inkling](/models/inkling) | 0.99 reasoning | Thinking Machines Lab | 2.08% |
| 49 | [MiMo-V2.5](/models/mimo-v2-5) | — | Xiaomi | 1.67% |
| 50 | [GPT-6 Sol](/models/gpt-6-sol) | max reasoning | OpenAI | 1.67% |
| 51 | [Kimi K2.6](/models/kimi-2-6) | — | Moonshot AI | 1.67% |
| 52 | [Inkling-Small](/models/inkling-small) | 0.99 reasoning | Thinking Machines Lab | 1.67% |
| 53 | [Qwen3.7 Max](/models/qwen3-7-max) | — | Alibaba | 1.67% |
| 54 | [Ling 3.0 Flash 2607](https://www.vals.ai/models/ant_ling-3.0-flash-2607) | — | Ant | 1.25% |
| 55 | [GPT-5.6 Luna](/models/gpt-5-6-luna) | max reasoning | OpenAI | 1.25% |
| 56 | [Qwen3.6 Plus](/models/qwen3-6-plus) | — | Alibaba | 1.25% |
| 57 | [Claude Haiku 5.5](/models/claude-haiku-5-5) | — | Anthropic | 1.25% |
| 58 | [Claude Haiku 4.5 Thinking](/models/claude-haiku-4-5-thinking) | — | Anthropic | 0.83% |
| 59 | [GPT-5.6 Terra](/models/gpt-5-6-terra) | max reasoning | OpenAI | 0.83% |
| 60 | [Grok 4.3](/models/grok-4-3) | high reasoning | xAI | 0.42% |
| 61 | [Nemotron 3 Ultra 550b A55b](https://www.vals.ai/models/nvidia_nemotron-3-ultra-550b-a55b) | — | Nvidia | 0.42% |
| 62 | [Mistral Medium 3.5](https://www.vals.ai/models/mistralai_mistral-medium-3.5) | high reasoning | Mistral AI | 0.42% |
| 63 | [Gemini 3.1 Flash Lite Preview](https://www.vals.ai/models/google_gemini-3.1-flash-lite-preview) | high reasoning | Google | 0.00% |
| 64 | [Mercury 2.5](/models/mercury-2-5) | high reasoning | Inception | 0.00% |
| 65 | [Gemini 3 Flash Preview](https://www.vals.ai/models/google_gemini-3-flash-preview) | high reasoning | Google | 0.00% |
| 66 | [Gemini 3.5 Flash-Lite](/models/gemini-3-5-flash-lite) | high reasoning | Google | 0.00% |
| 67 | [Ling 3.0 Flash Af Rc3](https://www.vals.ai/models/ant_ling-3.0-flash-af-rc3) | — | Ant | 0.00% |
| 68 | [GPT-5.4 nano](/models/gpt-5-4-nano) | high reasoning | OpenAI | 0.00% |
| 69 | [MiniMax M2.7](/models/minimax-m2-7) | — | MiniMax | 0.00% |
| 70 | [Gemini 3.1 Pro Preview](https://www.vals.ai/models/google_gemini-3.1-pro-preview) | high reasoning | Google | 0.00% |
| 71 | [Laguna M.1](/models/laguna-m-1) | — | Poolside | 0.00% |
| 72 | [GLM-5.1](/models/glm-5-1) | — | Z.AI | 0.00% |
| 73 | [Grok 4.20 0309 Reasoning](https://www.vals.ai/models/grok_grok-4.20-0309-reasoning) | — | xAI | 0.00% |
| 74 | [Qwen3.7 Plus](/models/qwen3-7-plus) | — | Alibaba | 0.00% |
| 75 | [GPT-5.4 mini](/models/gpt-5-4-mini) | xhigh reasoning | OpenAI | 0.00% |
| 76 | [GPT-5.4](/models/gpt-5-4) | xhigh reasoning | OpenAI | 0.00% |
| 77 | [Laguna XS.2](/models/laguna-xs-2) | — | Poolside | 0.00% |
| 78 | [Kimi K2.5 Thinking](https://www.vals.ai/models/kimi_kimi-k2.5-thinking) | — | Moonshot AI | 0.00% |
| 79 | [Nemotron Lightning 3p5 30b A3b](https://www.vals.ai/models/fireworks_nemotron-lightning-3p5-30b-a3b) | — | Fireworks AI | 0.00% |

## FAQ

### What does Harvey's Legal Agent Benchmark measure?

Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools

### Which model leads the published Harvey's Legal Agent Benchmark snapshot?

Muse Spark 1.2 currently leads the published Harvey's Legal Agent Benchmark snapshot with a score of 25.42%.

### How many models are evaluated on Harvey's Legal Agent Benchmark?

The October 12, 2026 contains 79 AI models.

### Does Harvey's Legal Agent Benchmark affect BenchLM's overall score?

Not directly. Harvey's Legal Agent Benchmark is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.