Benchmark Almanac

Legora BAR

Claude Opus 5.5 106.0 leads since Sep 2026 Not in model cards

Models work private legal cases end to end inside Legora's agent harness, with the matter's documents and templates. An LLM judge scores outputs against rubrics written by lawyers.

Cases span 30 practice areas and seven task families. Results are shown relative to the average of the models run, 17 models in the 2026-10 edition. Legora reruns it as models and harness change.

Made by Legora Released Sep 2026 Law Tasks kept private Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Graded by a model A language model scores each response, usually against a rubric.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Leaderboard legora.com
17
Models scored on its own leaderboard

Claude Opus 5.5 leads with 106.0

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

17 models. Scroll sideways for the rest.

  1. 106.0 Claude Opus 5.5
  2. 105.0 Muse Wonder
  3. 104.0 Claude Fable 5.1
  4. 103.0 Gemini 3.8 Flash
  5. 103.0 Mistral Large 4
  6. 102.0 Kimi K3
  7. 102.0 GLM-5.3
  8. 102.0 Grok 4.7
  9. 102.0 Claude Sonnet 5.5
  10. 102.0 GPT-6.1 Sol
  11. 101.0 Muse Spark 1.3
  12. 100.0 DeepSeek V4.1 Flash
  13. 99.0 Gemini 4 Argon
  14. 96.0 GPT-6 Astra
  15. 94.0 GLM-5.3-Flash
  16. 93.0 DeepSeek-V4-Pro
  17. 88.0 GPT-6 Luna
  1. 1 Claude Opus 5.5 106.0
  2. 2 Muse Wonder 105.0
  3. 3 Claude Fable 5.1 104.0
  4. 4 Gemini 3.8 Flash 103.0
  5. 5 Mistral Large 4 103.0
  6. 6 Kimi K3 102.0
  7. 7 GLM-5.3 102.0
  8. 8 Grok 4.7 102.0
  9. 9 Claude Sonnet 5.5 102.0
  10. 10 GPT-6.1 Sol 102.0
  11. 11 Muse Spark 1.3 101.0
  12. 12 DeepSeek V4.1 Flash 100.0
  13. 13 Gemini 4 Argon 99.0
  14. 14 GPT-6 Astra 96.0
  15. 15 GLM-5.3-Flash 94.0
  16. 16 DeepSeek-V4-Pro 93.0
  17. 17 GPT-6 Luna 88.0
#ModelScoreSettingDateSource
1 Claude Opus 5.5 106.0 Sep 2026 Official leaderboard
2 Muse Wonder 105.0 Oct 2026 Official leaderboard
3 Claude Fable 5.1 104.0 Sep 2026 Official leaderboard
4 Gemini 3.8 Flash 103.0 Sep 2026 Official leaderboard
5 Mistral Large 4 103.0 Oct 2026 Official leaderboard
6 Kimi K3 102.0 Jul 2026 Official leaderboard
7 GLM-5.3 102.0 Aug 2026 Official leaderboard
8 Grok 4.7 102.0 Sep 2026 Official leaderboard
9 Claude Sonnet 5.5 102.0 Sep 2026 Official leaderboard
10 GPT-6.1 Sol 102.0 Sep 2026 Official leaderboard
11 Muse Spark 1.3 101.0 Sep 2026 Official leaderboard
12 DeepSeek V4.1 Flash 100.0 Sep 2026 Official leaderboard
13 Gemini 4 Argon 99.0 Sep 2026 Official leaderboard
14 GPT-6 Astra 96.0 Sep 2026 Official leaderboard
15 GLM-5.3-Flash 94.0 Aug 2026 Official leaderboard
16 DeepSeek-V4-Pro 93.0 Apr 2026 Official leaderboard
17 GPT-6 Luna 88.0 Sep 2026 Official leaderboard

2 models have held number one since its release in Sep 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard 2 leaders
Claude Opus 5.5 106.0

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
85.0 90.0 95.0 100.0 105.0 110.0 2026 85.0 90.0 95.0 100.0 105.0 110.0 2026

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 140 in Law, finance and medicine
  • HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
  • MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy. Qwen 3.8 Max 80.4%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • KBL In use Models answer Korean legal knowledge and reasoning tasks and Korean bar exam questions. It has 7 knowledge tasks, 4 reasoning tasks and 2,510 bar exam items. Solar Pro 4 77.5%
  • KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. Solar Pro 4 93.2%
  • PRBench In use Model answers 1,100 expert-written law and finance questions. LLM judge grades against weighted expert rubrics of 10-30 criteria each. Muse Spark 1.3 61.6%