Benchmark Almanac

SnorkelManufacturing

Muse Wonder 27.8% leads since October 2026
Not in model cards

Agents make engineering, production, quality, supplier and safety decisions in a simulated discrete-parts plant across 200 tasks, a third of them CAD tasks.

Answer, state, safety, process and efficiency are scored separately.

Made by
Snorkel AI
Released
October 2026
Kind
Other professions
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

Muse Wonder leads with 27.8%

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. 27.8 Muse Wonder
  2. 19.6 GPT-6.1 Sol
  3. 19.5 Gemini 4 Argon
  4. 16.4 Grok 4.6
  5. 12.1 Grok 4.7
  6. 11.9 GLM-5.3
  7. 11.3 Claude Fable 5.1
  8. 10.5 Muse Spark 1.3
  9. 10.4 Claude Opus 5.5
  10. 10.2 Kimi K3
  11. 10.0 GPT-6 Astra
  12. 9.40 Claude Opus 5
  13. 8.30 Qwen 3.8 Max
  14. 8.10 DeepSeek-V4-Pro
  15. 7.10 Gemini 3.8 Flash
  16. 5.40 Nemotron 3 Ultra
  1. 1 Muse Wonder 27.8%
  2. 2 GPT-6.1 Sol 19.6%
  3. 3 Gemini 4 Argon 19.5%
  4. 4 Grok 4.6 16.4%
  5. 5 Grok 4.7 12.1%
  6. 6 GLM-5.3 11.9%
  7. 7 Claude Fable 5.1 11.3%
  8. 8 Muse Spark 1.3 10.5%
  9. 9 Claude Opus 5.5 10.4%
  10. 10 Kimi K3 10.2%
  11. 11 GPT-6 Astra 10%
  12. 12 Claude Opus 5 9.40%
  13. 13 Qwen 3.8 Max 8.30%
  14. 14 DeepSeek-V4-Pro 8.10%
  15. 15 Gemini 3.8 Flash 7.10%
  16. 16 Nemotron 3 Ultra 5.40%

Source: Official SnorkelManufacturing leaderboard

#ModelScoreSettingDateSource
1 Muse Wonder 27.8% Oct 2026 Official leaderboard
2 GPT-6.1 Sol 19.6% Sep 2026 Official leaderboard
3 Gemini 4 Argon 19.5% Sep 2026 Official leaderboard
4 Grok 4.6 16.4% Aug 2026 Official leaderboard
5 Grok 4.7 12.1% Sep 2026 Official leaderboard
6 GLM-5.3 11.9% Aug 2026 Official leaderboard
7 Claude Fable 5.1 11.3% Sep 2026 Official leaderboard
8 Muse Spark 1.3 10.5% Sep 2026 Official leaderboard
9 Claude Opus 5.5 10.4% Sep 2026 Official leaderboard
10 Kimi K3 10.2% Jul 2026 Official leaderboard
11 GPT-6 Astra 10% Sep 2026 Official leaderboard
12 Claude Opus 5 9.40% Jul 2026 Official leaderboard
13 Qwen 3.8 Max 8.30% Aug 2026 Official leaderboard
14 DeepSeek-V4-Pro 8.10% Apr 2026 Official leaderboard
15 Gemini 3.8 Flash 7.10% Sep 2026 Official leaderboard
16 Nemotron 3 Ultra 5.40% 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$4$6$8 Cost per trial ($), lower is better Pass@1 DeepSeek-V4-Pro: Pass@1 8.1%, cost per trial $1.08 Nemotron 3 Ultra: Pass@1 5.4%, cost per trial $0.5 Kimi K3: Pass@1 10.2%, cost per trial $2.62 Claude Opus 5: Pass@1 9.4%, cost per trial $3.12 Qwen 3.8 Max: Pass@1 8.3%, cost per trial $2.23 Grok 4.6: Pass@1 16.4%, cost per trial $1.98 GLM-5.3: Pass@1 11.9%, cost per trial $0.92 Claude Fable 5.1: Pass@1 11.3%, cost per trial $5.89 Gemini 3.8 Flash: Pass@1 7.1%, cost per trial $1.05 Muse Spark 1.3: Pass@1 10.5%, cost per trial $1.09 GPT-6 Astra: Pass@1 10%, cost per trial $7.39 Grok 4.7: Pass@1 12.1%, cost per trial $2.26 Claude Opus 5.5: Pass@1 10.4%, cost per trial $2.62 Gemini 4 Argon: Pass@1 19.5%, cost per trial $4.28 GPT-6.1 Sol: Pass@1 19.6%, cost per trial $0.25 Muse Wonder: Pass@1 27.8%, cost per trial $7.15 Muse WonderGPT-6.1 Sol 0% 10% 20% 30% $0$2$4$6$8 Cost per trial ($), lower is better Pass@1 DeepSeek-V4-Pro: Pass@1 8.1%, cost per trial $1.08 Nemotron 3 Ultra: Pass@1 5.4%, cost per trial $0.5 Kimi K3: Pass@1 10.2%, cost per trial $2.62 Claude Opus 5: Pass@1 9.4%, cost per trial $3.12 Qwen 3.8 Max: Pass@1 8.3%, cost per trial $2.23 Grok 4.6: Pass@1 16.4%, cost per trial $1.98 GLM-5.3: Pass@1 11.9%, cost per trial $0.92 Claude Fable 5.1: Pass@1 11.3%, cost per trial $5.89 Gemini 3.8 Flash: Pass@1 7.1%, cost per trial $1.05 Muse Spark 1.3: Pass@1 10.5%, cost per trial $1.09 GPT-6 Astra: Pass@1 10%, cost per trial $7.39 Grok 4.7: Pass@1 12.1%, cost per trial $2.26 Claude Opus 5.5: Pass@1 10.4%, cost per trial $2.62 Gemini 4 Argon: Pass@1 19.5%, cost per trial $4.28 GPT-6.1 Sol: Pass@1 19.6%, cost per trial $0.25 Muse Wonder: Pass@1 27.8%, cost per trial $7.15 Muse WonderGPT-6.1 Sol
ModelSettingPass@1Cost per trial
Muse Wonder 27.8% $7.15
GPT-6.1 Sol 19.6% $0.25
Gemini 4 Argon 19.5% $4.28
Grok 4.6 16.4% $1.98
Grok 4.7 12.1% $2.26
GLM-5.3 11.9% $0.92
Claude Fable 5.1 11.3% $5.89
Muse Spark 1.3 10.5% $1.09
Claude Opus 5.5 10.4% $2.62
Kimi K3 10.2% $2.62
GPT-6 Astra 10% $7.39
Claude Opus 5 9.4% $3.12
Qwen 3.8 Max 8.3% $2.23
DeepSeek-V4-Pro 8.1% $1.08
Gemini 3.8 Flash 7.1% $1.05
Nemotron 3 Ultra 5.4% $0.5

The best score, 27.8%, costs $7.15 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.

OpenAI Meta Cost frontier One model at each effort level
0% 10% 20% 30% $0.10 $0.20 $0.50 $1.00 $2.00 $5.00 $10 Cost per task, log scale Muse Wonder GPT-6.1 Sol 0% 10% 20% 30% $0.10 $1.00 $10 Cost per task, log scale Muse Wonder GPT-6.1 Sol

Source: Official SnorkelManufacturing leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Muse Wonder 27.8% $7.15
GPT-6.1 Sol 19.6% $0.25
Gemini 4 Argon 19.5% $4.28
Grok 4.6 16.4% $1.98
Grok 4.7 12.1% $2.26
GLM-5.3 11.9% $0.92
Claude Fable 5.1 11.3% $5.89
Muse Spark 1.3 10.5% $1.09
Claude Opus 5.5 10.4% $2.62
Kimi K3 10.2% $2.62
GPT-6 Astra 10% $7.39
Claude Opus 5 9.40% $3.12
Qwen 3.8 Max 8.30% $2.23
DeepSeek-V4-Pro 8.10% $1.08
Gemini 3.8 Flash 7.10% $1.05
Nemotron 3 Ultra 5.40% $0.50

2 models have held number one on the official leaderboard since its release in October 2026

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

OpenAI Meta
Own leaderboard 2 leaders
Muse Wonder 27.8%

Source: Official SnorkelManufacturing 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%) Muse Wonder GPT-6.1 Sol Kimi K3 DeepSeek-V4-Pro 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Muse Wonder GPT-6.1 Sol

Source: Official SnorkelManufacturing 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.

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%