Benchmark Almanac

SnorkelRevOps

Gemini 4 Argon 18.4% leads since October 2026
Not in model cards

Agents work 200 tasks in a simulated B2B SaaS revenue team spanning CRM, CPQ, billing, compensation, marketing and finance records.

State hashes verify required updates and rubrics grade memos and forecasts.

Made by
Snorkel AI
Released
October 2026
Kind
Running a business
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

Gemini 4 Argon leads with 18.4%

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. 18.4 Gemini 4 Argon
  2. 18.0 Muse Wonder
  3. 15.8 Grok 4.6
  4. 14.0 Claude Fable 5.1
  5. 13.1 Kimi K3
  6. 12.6 Muse Spark 1.3
  7. 12.5 Claude Opus 5.5
  8. 12.2 Claude Opus 5
  9. 12.1 GPT-6.1 Sol
  10. 11.8 GLM-5.3
  11. 11.3 Grok 4.7
  12. 10.9 GPT-6 Astra
  13. 10.1 Nemotron 3 Ultra
  14. 9.50 DeepSeek-V4-Pro
  15. 6.50 Qwen 3.8 Max
  16. 5.40 Gemini 3.8 Flash
  1. 1 Gemini 4 Argon 18.4%
  2. 2 Muse Wonder 18%
  3. 3 Grok 4.6 15.8%
  4. 4 Claude Fable 5.1 14%
  5. 5 Kimi K3 13.1%
  6. 6 Muse Spark 1.3 12.6%
  7. 7 Claude Opus 5.5 12.5%
  8. 8 Claude Opus 5 12.2%
  9. 9 GPT-6.1 Sol 12.1%
  10. 10 GLM-5.3 11.8%
  11. 11 Grok 4.7 11.3%
  12. 12 GPT-6 Astra 10.9%
  13. 13 Nemotron 3 Ultra 10.1%
  14. 14 DeepSeek-V4-Pro 9.50%
  15. 15 Qwen 3.8 Max 6.50%
  16. 16 Gemini 3.8 Flash 5.40%

Source: Official SnorkelRevOps leaderboard

#ModelScoreSettingDateSource
1 Gemini 4 Argon 18.4% Sep 2026 Official leaderboard
2 Muse Wonder 18% Oct 2026 Official leaderboard
3 Grok 4.6 15.8% Aug 2026 Official leaderboard
4 Claude Fable 5.1 14% Sep 2026 Official leaderboard
5 Kimi K3 13.1% Jul 2026 Official leaderboard
6 Muse Spark 1.3 12.6% Sep 2026 Official leaderboard
7 Claude Opus 5.5 12.5% Sep 2026 Official leaderboard
8 Claude Opus 5 12.2% Jul 2026 Official leaderboard
9 GPT-6.1 Sol 12.1% Sep 2026 Official leaderboard
10 GLM-5.3 11.8% Aug 2026 Official leaderboard
11 Grok 4.7 11.3% Sep 2026 Official leaderboard
12 GPT-6 Astra 10.9% Sep 2026 Official leaderboard
13 Nemotron 3 Ultra 10.1% Jun 2026 Official leaderboard
14 DeepSeek-V4-Pro 9.50% Apr 2026 Official leaderboard
15 Qwen 3.8 Max 6.50% Aug 2026 Official leaderboard
16 Gemini 3.8 Flash 5.40% Sep 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% 5% 10% 15% 20% $0$2$4$6$8$10 Cost per trial ($), lower is better Pass@1 DeepSeek-V4-Pro: Pass@1 9.5%, cost per trial $0.78 Nemotron 3 Ultra: Pass@1 10.1%, cost per trial $0.34 Claude Opus 5: Pass@1 12.2%, cost per trial $4.02 Qwen 3.8 Max: Pass@1 6.5%, cost per trial $1.87 GLM-5.3: Pass@1 11.8%, cost per trial $1.27 Claude Fable 5.1: Pass@1 14%, cost per trial $4.03 Gemini 3.8 Flash: Pass@1 5.4%, cost per trial $0.93 GPT-6 Astra: Pass@1 10.9%, cost per trial $5.92 Grok 4.7: Pass@1 11.3%, cost per trial $2.17 Claude Opus 5.5: Pass@1 12.5%, cost per trial $2.47 Muse Wonder: Pass@1 18%, cost per trial $8.38 Kimi K3: Pass@1 13.1%, cost per trial $1.77 Grok 4.6: Pass@1 15.8%, cost per trial $1.94 Muse Spark 1.3: Pass@1 12.6%, cost per trial $1.07 GPT-6.1 Sol: Pass@1 12.1%, cost per trial $0.18 Gemini 4 Argon: Pass@1 18.4%, cost per trial $3.75 Gemini 4 ArgonGrok 4.6Kimi K3Muse Spark 1.3GPT-6.1 Sol 0% 5% 10% 15% 20% $0$2$4$6$8$10 Cost per trial ($), lower is better Pass@1 DeepSeek-V4-Pro: Pass@1 9.5%, cost per trial $0.78 Nemotron 3 Ultra: Pass@1 10.1%, cost per trial $0.34 Claude Opus 5: Pass@1 12.2%, cost per trial $4.02 Qwen 3.8 Max: Pass@1 6.5%, cost per trial $1.87 GLM-5.3: Pass@1 11.8%, cost per trial $1.27 Claude Fable 5.1: Pass@1 14%, cost per trial $4.03 Gemini 3.8 Flash: Pass@1 5.4%, cost per trial $0.93 GPT-6 Astra: Pass@1 10.9%, cost per trial $5.92 Grok 4.7: Pass@1 11.3%, cost per trial $2.17 Claude Opus 5.5: Pass@1 12.5%, cost per trial $2.47 Muse Wonder: Pass@1 18%, cost per trial $8.38 Kimi K3: Pass@1 13.1%, cost per trial $1.77 Grok 4.6: Pass@1 15.8%, cost per trial $1.94 Muse Spark 1.3: Pass@1 12.6%, cost per trial $1.07 GPT-6.1 Sol: Pass@1 12.1%, cost per trial $0.18 Gemini 4 Argon: Pass@1 18.4%, cost per trial $3.75 Gemini 4 ArgonGrok 4.6Kimi K3Muse Spark 1.3GPT-6.1 Sol
ModelSettingPass@1Cost per trial
Gemini 4 Argon 18.4% $3.75
Muse Wonder 18% $8.38
Grok 4.6 15.8% $1.94
Claude Fable 5.1 14% $4.03
Kimi K3 13.1% $1.77
Muse Spark 1.3 12.6% $1.07
Claude Opus 5.5 12.5% $2.47
Claude Opus 5 12.2% $4.02
GPT-6.1 Sol 12.1% $0.18
GLM-5.3 11.8% $1.27
Grok 4.7 11.3% $2.17
GPT-6 Astra 10.9% $5.92
Nemotron 3 Ultra 10.1% $0.34
DeepSeek-V4-Pro 9.5% $0.78
Qwen 3.8 Max 6.5% $1.87
Gemini 3.8 Flash 5.4% $0.93

The best score, 18.4%, costs $3.75 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.

Moonshot AI xAI Meta OpenAI Google DeepMind Cost frontier One model at each effort level
0% 5% 10% 15% 20% $0.10 $0.20 $0.50 $1.00 $2.00 $5.00 $10 Cost per task, log scale Gemini 4 Argon Grok 4.6 Kimi K3 Muse Spark 1.3 GPT-6.1 Sol 0% 5% 10% 15% 20% $0.10 $1.00 $10 Cost per task, log scale Gemini 4 Argon Grok 4.6

Source: Official SnorkelRevOps leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Argon 18.4% $3.75
Muse Wonder 18% $8.38
Grok 4.6 15.8% $1.94
Claude Fable 5.1 14% $4.03
Kimi K3 13.1% $1.77
Muse Spark 1.3 12.6% $1.07
Claude Opus 5.5 12.5% $2.47
Claude Opus 5 12.2% $4.02
GPT-6.1 Sol 12.1% $0.18
GLM-5.3 11.8% $1.27
Grok 4.7 11.3% $2.17
GPT-6 Astra 10.9% $5.92
Nemotron 3 Ultra 10.1% $0.34
DeepSeek-V4-Pro 9.50% $0.78
Qwen 3.8 Max 6.50% $1.87
Gemini 3.8 Flash 5.40% $0.93

Number one over time

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

Google DeepMind
Own leaderboard 1 leader
Gemini 4 Argon 18.4%

Source: Official SnorkelRevOps 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%) Gemini 4 Argon Kimi K3 DeepSeek-V4-Pro 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Gemini 4 Argon Nemotron 3 Ultra

Source: Official SnorkelRevOps 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 Economic work

All 97 in Economic work
  • JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
  • GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate. Seed2.1 Pro 87.9%
  • GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
  • AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
  • CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%
  • OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%