Benchmark Almanac

Contract Workflows (Models)

Claude Fable 5.1 48.4% leads since September 2026
Not in model cards

Systems complete legal work involving contracts. The benchmark assesses whether each deliverable satisfies its legal and presentation criteria.

The publisher reports results for models under its track methodology.

Kind
Law
Level
Expert
Scored
Graded by people
Answers
Open-ended
Measures
Capability
Subject
Law
Language
English
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
A host
In views
Law and finance
Strongest model tested
Claude Fable 5.2 #2 on the capability index
Last model added
November 2026
Scores refreshed
November 20th, 2026
27
Models scored on its own leaderboard

3 models share the top score of 48.4%

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

  1. 1 Claude Fable 5.1 48.4%
  2. 2 Muse Spark 1.3 48.4%
  3. 3 Claude Opus 5.5 48.4%
  4. 4 Claude Opus 5 45.2%
  5. 5 Qwen3.8 Flash 38.7%
  6. 6 Claude Haiku 5.5 35.5%
  7. 7 Claude Fable 5.2 32.3%
  8. 8 Qwen 3.8 Max 29%
  9. 9 Claude Sonnet 5.5 29%
  10. 10 Grok 4.6 25.8%
  11. 11 Gemini 3.8 Flash 25.8%
  12. 12 GPT-6 Astra 25.8%
  13. 13 GPT-6.1 Sol 25.8%
  14. 14 Gemini 4 Argon 25.8%
  15. 15 Mistral Large 4 25.8%
  16. 16 DeepSeek-V4-Flash 22.6%
  17. 17 GPT-5.6 Sol 22.6%
  18. 18 Kimi K3 22.6%
  19. 19 GLM-5.3-Flash 22.6%
  20. 20 DeepSeek-V4-Pro 19.4%
  21. 21 GPT-5.6 Luna 19.4%
  22. 22 GLM-5.3 19.4%
  23. 23 Claude Sonnet 5 16.1%
  24. 24 GPT-5.6 Terra 16.1%
  25. 25 MiniMax-M3 12.9%
  26. 26 GPT-6.1 Terra 12.9%
  27. 27 Mistral Medium 3.5 9.70%

Source: Official Contract Workflows (Models) leaderboard

#ModelScoreSettingDateSource
1 Claude Fable 5.1 48.4% Sep 2026 Official leaderboard
2 Muse Spark 1.3 48.4% Sep 2026 Official leaderboard
3 Claude Opus 5.5 48.4% Sep 2026 Official leaderboard
4 Claude Opus 5 45.2% Jul 2026 Official leaderboard
5 Qwen3.8 Flash 38.7% Aug 2026 Official leaderboard
6 Claude Haiku 5.5 35.5% Oct 2026 Official leaderboard
7 Claude Fable 5.2 32.3% Nov 2026 Official leaderboard
8 Qwen 3.8 Max 29% Aug 2026 Official leaderboard
9 Claude Sonnet 5.5 29% Sep 2026 Official leaderboard
10 Grok 4.6 25.8% Aug 2026 Official leaderboard
11 Gemini 3.8 Flash 25.8% Sep 2026 Official leaderboard
12 GPT-6 Astra 25.8% Sep 2026 Official leaderboard
13 GPT-6.1 Sol 25.8% Sep 2026 Official leaderboard
14 Gemini 4 Argon 25.8% Sep 2026 Official leaderboard
15 Mistral Large 4 25.8% Oct 2026 Official leaderboard
16 DeepSeek-V4-Flash 22.6% Apr 2026 Official leaderboard
17 GPT-5.6 Sol 22.6% Jul 2026 Official leaderboard
18 Kimi K3 22.6% Jul 2026 Official leaderboard
19 GLM-5.3-Flash 22.6% Aug 2026 Official leaderboard
20 DeepSeek-V4-Pro 19.4% Apr 2026 Official leaderboard
21 GPT-5.6 Luna 19.4% Jul 2026 Official leaderboard
22 GLM-5.3 19.4% Aug 2026 Official leaderboard
23 Claude Sonnet 5 16.1% Jun 2026 Official leaderboard
24 GPT-5.6 Terra 16.1% Jul 2026 Official leaderboard
25 MiniMax-M3 12.9% Jun 2026 Official leaderboard
26 GPT-6.1 Terra 12.9% Oct 2026 Official leaderboard
27 Mistral Medium 3.5 9.70% Apr 2026 Official leaderboard

The best score, 48.4%, costs $0.20 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

DeepSeek Qwen Meta Cost frontier One model at each effort level
0% 10% 20% 30% 40% 50% $0.01 $0.10 $1.00 $10 Cost per task, log scale Muse Spark 1.3 Qwen3.8 Flash DeepSeek-V4-Flash 0% 10% 20% 30% 40% 50% $0.01 $0.10 $1.00 $10 Cost per task, log scale Muse Spark 1.3 Qwen3.8 Flash

Source: Official Contract Workflows (Models) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Fable 5.1 48.4% $1.50
Muse Spark 1.3 48.4% $0.20
Claude Opus 5.5 48.4% $1.47
Claude Opus 5 45.2% $0.90
Qwen3.8 Flash 38.7% $0.03
Claude Haiku 5.5 35.5% $0.16
Claude Fable 5.2 32.3% $2.98
Qwen 3.8 Max 29% $0.39
Claude Sonnet 5.5 29% $0.98
Grok 4.6 25.8% $0.28
Gemini 3.8 Flash 25.8% $0.22
GPT-6 Astra 25.8% $1.13
GPT-6.1 Sol 25.8% $0.11
Gemini 4 Argon 25.8% $0.26
Mistral Large 4 25.8% $0.09
DeepSeek-V4-Flash 22.6% $0.01
GPT-5.6 Sol 22.6% $0.18
Kimi K3 22.6% $0.26
GLM-5.3-Flash 22.6% $0.02
DeepSeek-V4-Pro 19.4% $0.08
GPT-5.6 Luna 19.4% $0.02
GLM-5.3 19.4% $0.09
Claude Sonnet 5 16.1% $0.36
GPT-5.6 Terra 16.1% $0.14
MiniMax-M3 12.9% $0.03
GPT-6.1 Terra 12.9% $0.07
Mistral Medium 3.5 9.70% $0.32

3 models have held number one on the official leaderboard since 2023

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

DeepSeek Anthropic
Own leaderboard 3 leaders

Source: Official Contract Workflows (Models) leaderboard

Scores over time

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

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Fable 5.1 DeepSeek-V4-Flash 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Fable 5.1 DeepSeek-V4-Flash

Source: Official Contract Workflows (Models) leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 637 in Law, finance and medicine
  • 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 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%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy. Ling-3.1-flash 81.7%
  • 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. o1-preview 92.7%
  • Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%

Data audit

10 of 11 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 5 pass
  • Pass Independent runs Scores from the official leaderboard (27 entries).
  • Pass Models scored 27 models we track.
  • Pass Newest score The newest is from November 2026, for Claude Fable 5.2.
  • Pass Frontier models 8 of the top 10 have a score, led by Claude Fable 5.2 (#2).
  • Pass Headroom The best is 48.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on legalbenchmarks.ai: 27 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check