BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7

Legal Research Bench leaderboard

As of 6 Nov 2026, the top score on Legal Research Bench is 55.8% by Gemini 4 Carbon (Google), ahead of Claude Fable 5.1 at 55.3%, across 88 model configurations with a published result (Vals AI, result dated 5 Nov 2026). The model must research a US legal question with tools and return a supported answer, the way an associate would. Legal Research Bench is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 6 Nov 2026)

  • Gemini 4 Carbon holds the top Legal Research Bench score at 55.8% (Vals AI, 5 Nov 2026).
  • Claude Fable 5.1 is second on Legal Research Bench at 55.3% (Vals AI, 5 Nov 2026).
  • Legal Research Bench: Agentic US legal research questions. Run by Vals AI (Vals AI).
  • 88 model configurations have a published Legal Research Bench result on BenchLeader (BenchLeader, 6 Nov 2026).
  • Legal Research Bench does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Agentic US legal research questions. Run by Vals AI.

Published by
Vals AI
Category
Agents & tools
Index weight
Reference only
Models
88
Data as of
6 Nov 2026

Vals AI (vals.ai).

What the test looks like

The model must research a US legal question with tools and return a supported answer, the way an associate would.

How it is scored

Accuracy graded by lawyers, run by Vals AI.

What to keep in mind

US-specific and dependent on the research tools provided.

Top 15
  1. 1Gemini 4 Carbon (high)55.8%
  2. 2Claude Fable 5.155.3%
  3. 3Claude Opus 555.3%
  4. 4Muse Spark 1.3 (max)55.3%
  5. 5Gemini 4 Argon (high)54.8%
  6. 6Muse Wonder Full (max)53.4%
  7. 7Claude Opus 5.550.5%
  8. 8Muse Wonder (max)50.5%
  9. 9Claude Fable 549.5%
  10. 10GLM 5.3 (max)49.0%
  11. 11GPT-5.6 Sol (max)48.1%
  12. 12Grok 4.6 (high)48.1%
  13. 13Claude Sonnet 5.548.1%
  14. 14Qwen3.8 Max (max)47.6%
  15. 15Grok 4.7 (xhigh)47.1%
88 of 88
#
1Gemini 4 Carbonhigh55.8%69.2$8.742026-11-05
2Claude Fable 5.155.3%66.4$23.062026-11-05
3Claude Opus 555.3%65.2$6.762026-11-05
4Muse Spark 1.3max55.3%62.0$0.5892026-11-05
5Gemini 4 Argonhigh54.8%66.0$6.452026-11-05
6Muse Wonder Fullmax53.4%62.9$3.602026-11-05
7Claude Opus 5.550.5%63.4$25.162026-11-05
8Muse Wondermax50.5%62.3$3.982026-11-05
9Claude Fable 549.5%66.5$9.792026-11-05
10GLM 5.3max49.0%59.0$2.242026-11-05
11GPT-5.6 Solmax48.1%65.4$21.612026-11-05
12Grok 4.6high48.1%60.5$1.532026-11-05
13Claude Sonnet 5.548.1%–$14.832026-11-05
14Qwen3.8 Maxmax47.6%58.8$2.492026-11-05
15Grok 4.7xhigh47.1%58.3$4.922026-11-05
16MiMo-V2.6-Pro47.1%57.3$0.1812026-11-05
17NewInkling 2 Small46.6%57.4$0.2892026-11-05
18Kimi K3max46.1%62.1$3.472026-11-05
19Qwen3.8 Max Internal46.1%–$5.312026-11-05
20DeepSeek V4.1 Flash 1017max45.2%59.2$0.6992026-11-05
21Hy445.2%55.7$0.8462026-11-05
22GLM 5.3 Flashmax45.2%54.2$0.2402026-11-05
23Ember-144.7%–$3.672026-11-05
24Muse Spark 1.2xhigh43.8%61.7$0.5082026-11-05
25Claude Opus 4.843.8%60.0$2.822026-11-05
26Claude Haiku 5.543.8%–$1.652026-11-05
27Claude Sonnet 541.8%55.9$2.722026-11-05
28GLM-Legalmax41.8%–$0.0492026-11-05
29GPT-5.6 Terramax41.4%60.7$7.912026-11-05
30DeepSeek V4.1 Flashmax41.4%59.0$0.2482026-11-05
31Muse Spark 1.3xhigh40.9%58.5$0.6502026-11-05
32DeepSeek V4 Promax40.9%57.0$1.092026-11-05
33GPT-5.5xhigh40.4%64.3$7.402026-11-05
34Step 5 Preview40.4%54.9$0.8392026-11-05
35GPT-6 Astramax39.4%67.8$10.582026-11-05
36Gemini 3.8 Flashhigh38.9%62.0$1.632026-11-05
37GPT-6.1 Solmax38.5%67.3$3.382026-11-05
38Claude Opus 4.738.5%63.1$3.062026-11-05
39Claude Sonnet 4.638.5%58.2$2.302026-11-05
40Muse Spark Openxhigh38.5%54.0$1.562026-11-05
41Muse Spark 1.1xhigh38.0%59.5$0.3772026-11-05
42Grok 4.5high38.0%56.9$1.092026-11-05
43MiMo-V2.6-Flash38.0%55.0$0.0702026-11-05
44Kimi K3.1 Promax37.5%–$7.232026-11-05
45GPT-5.6 Lunamax36.5%56.3$0.8452026-11-05
46Qwen3.8 27Bxhigh36.1%52.5$2.162026-11-05
47Mistral Large 4 OSShigh35.6%–$1.092026-11-05
48Gemini 3.7 Flashhigh34.6%62.2$0.7602026-11-05
49MiniMax M3.1 Flash34.1%53.0$0.4452026-11-05
50Mistral Large 4high31.7%49.5$2.722026-11-05
51GLM 5.2max31.3%59.5$0.8892026-11-05
52Gemini 3.5 Flashhigh30.8%62.5$1.312026-11-05
53DeepSeek V4 Flashhigh30.3%57.3$0.2752026-11-05
54GPT-6 Lunamax30.3%56.7$0.4402026-11-05
55MiniMax M329.8%56.4$0.3432026-11-05
56GPT-6 Solmax28.9%63.3$4.902026-11-05
57Beam 1.028.9%45.8–2026-11-05
58Mercury 3high28.9%–$0.0742026-11-05
59Inkling28.4%47.5$0.4942026-11-05
60GLM 5.127.9%55.7$0.3862026-11-05

More agents & tools benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on Legal Research Bench

Cite as: BenchLeader, “Legal Research Bench leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/vals_legal_research.

Legal Research Bench: questions

What does Legal Research Bench measure?
The model must research a US legal question with tools and return a supported answer, the way an associate would. Scores are reported in percent of tasks solved; higher is better.
What is the top score on Legal Research Bench?
As of 6 Nov 2026, the highest score on Legal Research Bench is 55.8% by Gemini 4 Carbon (result dated 5 Nov 2026), ahead of Claude Fable 5.1 at 55.3%. Scores come from Vals AI and are re-read daily.
Which AI model leads Legal Research Bench?
Gemini 4 Carbon leads Legal Research Bench with 55.8% as of 6 Nov 2026 (result dated 5 Nov 2026), ahead of Claude Fable 5.1 at 55.3%.
How many models have Legal Research Bench results?
88 model configurations have a Legal Research Bench result on BenchLeader, all taken from Vals AI.
Who runs Legal Research Bench and how often is it updated?
Legal Research Bench is published by Vals AI. BenchLeader re-reads the published results once a day and records the date each result was published.
Does Legal Research Bench count toward the BenchLeader Index?
No. Legal Research Bench is shown for reference but left out of the composite index.