Benchmark Almanac

SnorkelFinance 2.0

Grok 4.6 25.1% leads since October 2026
Not in model cards

Agents do 200 investment banking and private equity workflow tasks in a simulated firm with incomplete requests, scattered evidence, compliance rules and required state changes.

Exact outputs and state are checked by code and semantic answers by rubric. The board reports pass@1. Part of Snorkel Enterprise Environments.

Made by
Snorkel AI
Released
October 2026
Kind
Finance
Level
Expert
Scored
Graded by a model
Answers
Agent task
Tasks
Kept private
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
16
Models scored on its own leaderboard

Grok 4.6 leads with 25.1%

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

16 models. Scroll sideways for the rest.

  1. 25.1 Grok 4.6
  2. 21.2 GLM-5.3
  3. 20.2 Grok 4.7
  4. 19.6 Claude Fable 5.1
  5. 18.6 Claude Opus 5.5
  6. 17.3 Kimi K3
  7. 16.5 Claude Opus 5
  8. 15.1 Gemini 4 Argon
  9. 15.0 Muse Wonder
  10. 14.1 Qwen 3.8 Max
  11. 11.6 GPT-6 Astra
  12. 10.0 Muse Spark 1.3
  13. 9.50 DeepSeek-V4-Pro
  14. 6.60 GPT-6.1 Sol
  15. 5.40 Gemini 3.8 Flash
  16. 4.30 Nemotron 3 Ultra
  1. 1 Grok 4.6 25.1%
  2. 2 GLM-5.3 21.2%
  3. 3 Grok 4.7 20.2%
  4. 4 Claude Fable 5.1 19.6%
  5. 5 Claude Opus 5.5 18.6%
  6. 6 Kimi K3 17.3%
  7. 7 Claude Opus 5 16.5%
  8. 8 Gemini 4 Argon 15.1%
  9. 9 Muse Wonder 15%
  10. 10 Qwen 3.8 Max 14.1%
  11. 11 GPT-6 Astra 11.6%
  12. 12 Muse Spark 1.3 10%
  13. 13 DeepSeek-V4-Pro 9.50%
  14. 14 GPT-6.1 Sol 6.60%
  15. 15 Gemini 3.8 Flash 5.40%
  16. 16 Nemotron 3 Ultra 4.30%

Source: Official SnorkelFinance 2.0 leaderboard

#ModelScoreSettingDateSource
1 Grok 4.6 25.1% Aug 2026 Official leaderboard
2 GLM-5.3 21.2% Aug 2026 Official leaderboard
3 Grok 4.7 20.2% Sep 2026 Official leaderboard
4 Claude Fable 5.1 19.6% Sep 2026 Official leaderboard
5 Claude Opus 5.5 18.6% Sep 2026 Official leaderboard
6 Kimi K3 17.3% Jul 2026 Official leaderboard
7 Claude Opus 5 16.5% Jul 2026 Official leaderboard
8 Gemini 4 Argon 15.1% Sep 2026 Official leaderboard
9 Muse Wonder 15% Oct 2026 Official leaderboard
10 Qwen 3.8 Max 14.1% Aug 2026 Official leaderboard
11 GPT-6 Astra 11.6% Sep 2026 Official leaderboard
12 Muse Spark 1.3 10% Sep 2026 Official leaderboard
13 DeepSeek-V4-Pro 9.50% Apr 2026 Official leaderboard
14 GPT-6.1 Sol 6.60% Sep 2026 Official leaderboard
15 Gemini 3.8 Flash 5.40% Sep 2026 Official leaderboard
16 Nemotron 3 Ultra 4.30% Jun 2026 Official leaderboard

Pass@1 vs. cost per trial

The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

0% 10% 20% 30% $0$2.5$5$7.5$10$12.5 Cost per trial ($), lower is better Pass@1 DeepSeek-V4-Pro: Pass@1 9.5%, cost per trial $1.25 Kimi K3: Pass@1 17.3%, cost per trial $1.78 Claude Opus 5: Pass@1 16.5%, cost per trial $3.2 Qwen 3.8 Max: Pass@1 14.1%, cost per trial $1.71 Claude Fable 5.1: Pass@1 19.6%, cost per trial $5.83 Gemini 3.8 Flash: Pass@1 5.4%, cost per trial $1.11 Muse Spark 1.3: Pass@1 10%, cost per trial $0.98 GPT-6 Astra: Pass@1 11.6%, cost per trial $10.6 Grok 4.7: Pass@1 20.2%, cost per trial $2.04 Claude Opus 5.5: Pass@1 18.6%, cost per trial $4.01 GPT-6.1 Sol: Pass@1 6.6%, cost per trial $3.41 Gemini 4 Argon: Pass@1 15.1%, cost per trial $9.8 Muse Wonder: Pass@1 15%, cost per trial $7.4 Nemotron 3 Ultra: Pass@1 4.3%, cost per trial $0.42 Grok 4.6: Pass@1 25.1%, cost per trial $3.46 GLM-5.3: Pass@1 21.2%, cost per trial $0.88 Grok 4.6GLM-5.3Nemotron 3 Ultra 0% 10% 20% 30% $0$2.5$5$7.5$10$12.5 Cost per trial ($), lower is better Pass@1 DeepSeek-V4-Pro: Pass@1 9.5%, cost per trial $1.25 Kimi K3: Pass@1 17.3%, cost per trial $1.78 Claude Opus 5: Pass@1 16.5%, cost per trial $3.2 Qwen 3.8 Max: Pass@1 14.1%, cost per trial $1.71 Claude Fable 5.1: Pass@1 19.6%, cost per trial $5.83 Gemini 3.8 Flash: Pass@1 5.4%, cost per trial $1.11 Muse Spark 1.3: Pass@1 10%, cost per trial $0.98 GPT-6 Astra: Pass@1 11.6%, cost per trial $10.6 Grok 4.7: Pass@1 20.2%, cost per trial $2.04 Claude Opus 5.5: Pass@1 18.6%, cost per trial $4.01 GPT-6.1 Sol: Pass@1 6.6%, cost per trial $3.41 Gemini 4 Argon: Pass@1 15.1%, cost per trial $9.8 Muse Wonder: Pass@1 15%, cost per trial $7.4 Nemotron 3 Ultra: Pass@1 4.3%, cost per trial $0.42 Grok 4.6: Pass@1 25.1%, cost per trial $3.46 GLM-5.3: Pass@1 21.2%, cost per trial $0.88 Grok 4.6GLM-5.3Nemotron 3 Ultra
ModelSettingPass@1Cost per trial
Grok 4.6 25.1% $3.46
GLM-5.3 21.2% $0.88
Grok 4.7 20.2% $2.04
Claude Fable 5.1 19.6% $5.83
Claude Opus 5.5 18.6% $4.01
Kimi K3 17.3% $1.78
Claude Opus 5 16.5% $3.2
Gemini 4 Argon 15.1% $9.8
Muse Wonder 15% $7.4
Qwen 3.8 Max 14.1% $1.71
GPT-6 Astra 11.6% $10.6
Muse Spark 1.3 10% $0.98
DeepSeek-V4-Pro 9.5% $1.25
GPT-6.1 Sol 6.6% $3.41
Gemini 3.8 Flash 5.4% $1.11
Nemotron 3 Ultra 4.3% $0.42

The best score, 25.1%, costs $3.46 a task

Runs on the cost frontier score higher than any run that's cheaper. GLM-5.3 comes within 3.9 points of the best for $0.88 a task, 4 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

NVIDIA xAI Z.ai Cost frontier One model at each effort level
0% 10% 20% 30% $0.10 $1.00 $10 $100 Cost per task, log scale Grok 4.6 GLM-5.3 Nemotron 3 Ultra 0% 10% 20% 30% $0.10 $1.00 $10 $100 Cost per task, log scale Grok 4.6 GLM-5.3

Source: Official SnorkelFinance 2.0 leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Grok 4.6 25.1% $3.46
GLM-5.3 21.2% $0.88
Grok 4.7 20.2% $2.04
Claude Fable 5.1 19.6% $5.83
Claude Opus 5.5 18.6% $4.01
Kimi K3 17.3% $1.78
Claude Opus 5 16.5% $3.20
Gemini 4 Argon 15.1% $9.80
Muse Wonder 15% $7.40
Qwen 3.8 Max 14.1% $1.71
GPT-6 Astra 11.6% $11
Muse Spark 1.3 10% $0.98
DeepSeek-V4-Pro 9.50% $1.25
GPT-6.1 Sol 6.60% $3.41
Gemini 3.8 Flash 5.40% $1.11
Nemotron 3 Ultra 4.30% $0.42

Number one over time

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

xAI
Own leaderboard 1 leader
Grok 4.6 25.1%

Source: Official SnorkelFinance 2.0 leaderboard

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 Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Grok 4.6 Kimi K3 DeepSeek-V4-Pro 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Grok 4.6 Kimi K3

Source: Official SnorkelFinance 2.0 leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

Amazon OpenAI Anthropic xAI
SnorkelFinance 2.0 5 leaders
Grok 4.6 25.1%
SnorkelFinance Not tested since Nov 2025
o3 81%
SnorkelFinance 2.0 1 leader
Grok 4.6 25.1%

Source: Official leaderboards

Replaces SnorkelFinance

More in Law, finance and medicine

All 179 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%
  • 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%