Benchmark Almanac

Excel Modeling Benchmark v1.1

Claude Fable 5.1 76.7% leads since Oct 2026 Not in model cards

Agents build the same 103 Excel financial modeling tasks as EMB v1.0, with the Scratch rubric judge changed to GPT-6 Luna.

Vals says v1.1 scores are not directly comparable with v1.0. Part of Vals Index v2.2 from 2026-10-19. Claude Fable 5.1 led 37 models at 76.73% on 2026-10-29.

Made by Vals AI Released Oct 2026 Finance 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
37
Models scored on its own leaderboard

Claude Fable 5.1 leads with 76.7%

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. Every run is in the next chart.

37 models. Scroll sideways for the rest.

  1. 76.7 Claude Fable 5.1
  2. 76.5 Gemini 4 Carbon
  3. 75.9 Claude Sonnet 5.5
  4. 75.8 Claude Opus 5.5
  5. 75.0 Muse Wonder Full
  6. 74.8 Gemini 4 Argon
  7. 73.3 GPT-6.1 Terra
  8. 72.9 Gemini 3.8 Flash
  9. 72.7 GPT-6 Astra
  10. 72.5 Muse Wonder
  11. 72.0 Qwen3.8 Max (0902)
  12. 71.5 Grok 4.7
  13. 71.3 Kimi K3
  14. 71.1 GPT-6.1 Sol
  15. 70.0 Kimi K3.1 Pro
  16. 70.0 GPT-6 Luna
  17. 69.5 Step 5 Preview
  18. 69.2 DeepSeek V4.1 Flash 1017
  19. 69.1 GLM-5.3
  20. 69.0 Mistral Large 4
  21. 69.0 Tencent Hy4 preview
  22. 68.9 MiMo-V2.6 Flash
  23. 68.7 Mistral Large 4 (Le Chonk, open weights)
  24. 67.8 MiniMax-M3.1-Flash
  25. 67.8 DeepSeek V4.1 Flash
  26. 67.3 MiMo-V2.6 Pro
  27. 66.3 GLM-5.3-Flash
  28. 66.0 Qwen 3.8 Max
  29. 65.5 Claude Haiku 5.5
  30. 65.1 DeepSeek V4 Pro 0813
  31. 64.7 Beam 1.0
  32. 62.0 Muse Spark Open
  33. 59.8 MiniMax-M3
  34. 57.9 Mercury 3
  35. 52.8 Nemotron 3.5 Super VL
  36. 41.0 Inkling
  37. 40.8 Inkling-Small
  1. 1 Claude Fable 5.1 76.7%
  2. 2 Gemini 4 Carbon 76.5%
  3. 3 Claude Sonnet 5.5 75.9%
  4. 4 Claude Opus 5.5 75.8%
  5. 5 Muse Wonder Full 75%
  6. 6 Gemini 4 Argon 74.8%
  7. 7 GPT-6.1 Terra 73.3%
  8. 8 Gemini 3.8 Flash 72.9%
  9. 9 GPT-6 Astra 72.7%
  10. 10 Muse Wonder 72.5%
  11. 11 Qwen3.8 Max (0902) 72%
  12. 12 Grok 4.7 71.5%
  13. 13 Kimi K3 71.3%
  14. 14 GPT-6.1 Sol 71.1%
  15. 15 Kimi K3.1 Pro 70%
  16. 16 GPT-6 Luna 70%
  17. 17 Step 5 Preview 69.5%
  18. 18 DeepSeek V4.1 Flash 1017 69.2%
  19. 19 GLM-5.3 69.1%
  20. 20 Mistral Large 4 69%
  21. 21 Tencent Hy4 preview 69%
  22. 22 MiMo-V2.6 Flash 68.9%
  23. 23 Mistral Large 4 (Le Chonk, open weights) 68.7%
  24. 24 MiniMax-M3.1-Flash 67.8%
  25. 25 DeepSeek V4.1 Flash 67.8%
  26. 26 MiMo-V2.6 Pro 67.3%
  27. 27 GLM-5.3-Flash 66.3%
  28. 28 Qwen 3.8 Max 66%
  29. 29 Claude Haiku 5.5 65.5%
  30. 30 DeepSeek V4 Pro 0813 65.1%
  31. 31 Beam 1.0 64.7%
  32. 32 Muse Spark Open 62%
  33. 33 MiniMax-M3 59.8%
  34. 34 Mercury 3 57.9%
  35. 35 Nemotron 3.5 Super VL 52.8%
  36. 36 Inkling 41%
  37. 37 Inkling-Small 40.8%
#ModelScoreSettingDateSource
1 Claude Fable 5.1 76.7% Sep 2026 Official leaderboard
2 Gemini 4 Carbon 76.5% Oct 2026 Official leaderboard
3 Claude Sonnet 5.5 75.9% Sep 2026 Official leaderboard
4 Claude Opus 5.5 75.8% Sep 2026 Official leaderboard
5 Muse Wonder Full 75% Oct 2026 Official leaderboard
6 Gemini 4 Argon 74.8% Sep 2026 Official leaderboard
7 GPT-6.1 Terra 73.3% Oct 2026 Official leaderboard
8 Gemini 3.8 Flash 72.9% Sep 2026 Official leaderboard
9 GPT-6 Astra 72.7% Sep 2026 Official leaderboard
10 Muse Wonder 72.5% Oct 2026 Official leaderboard
11 Qwen3.8 Max (0902) 72% Sep 2026 Official leaderboard
12 Grok 4.7 71.5% Sep 2026 Official leaderboard
13 Kimi K3 71.3% Jul 2026 Official leaderboard
14 GPT-6.1 Sol 71.1% Sep 2026 Official leaderboard
15 Kimi K3.1 Pro 70% Oct 2026 Official leaderboard
16 GPT-6 Luna 70% Sep 2026 Official leaderboard
17 Step 5 Preview 69.5% Sep 2026 Official leaderboard
18 DeepSeek V4.1 Flash 1017 69.2% Oct 2026 Official leaderboard
19 GLM-5.3 69.1% Aug 2026 Official leaderboard
20 Mistral Large 4 69% Oct 2026 Official leaderboard
21 Tencent Hy4 preview 69% Aug 2026 Official leaderboard
22 MiMo-V2.6 Flash 68.9% Sep 2026 Official leaderboard
23 Mistral Large 4 (Le Chonk, open weights) 68.7% Oct 2026 Official leaderboard
24 MiniMax-M3.1-Flash 67.8% Oct 2026 Official leaderboard
25 DeepSeek V4.1 Flash 67.8% Sep 2026 Official leaderboard
26 MiMo-V2.6 Pro 67.3% Sep 2026 Official leaderboard
27 GLM-5.3-Flash 66.3% Aug 2026 Official leaderboard
28 Qwen 3.8 Max 66% Aug 2026 Official leaderboard
29 Claude Haiku 5.5 65.5% Oct 2026 Official leaderboard
30 DeepSeek V4 Pro 0813 65.1% Aug 2026 Official leaderboard
31 Beam 1.0 64.7% Oct 2026 Official leaderboard
32 Muse Spark Open 62% Oct 2026 Official leaderboard
33 MiniMax-M3 59.8% Jun 2026 Official leaderboard
34 Mercury 3 57.9% Oct 2026 Official leaderboard
35 Nemotron 3.5 Super VL 52.8% Oct 2026 Official leaderboard
36 Inkling 41% Jul 2026 Official leaderboard
37 Inkling-Small 40.8% Jul 2026 Official leaderboard

The best score, 76.7%, costs $16 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-6.1 Terra comes within 3.4 points of the best for $0.56 a task, 28 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Anthropic Xiaomi Corp OpenAI Meta Google DeepMind Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.10 $1.00 $10 $100 Cost per task, log scale Claude Fable 5.1 Gemini 4 Carbon GPT-6.1 Terra GPT-6 Luna 0% 20% 40% 60% 80% 100% $0.10 $1.00 $10 $100 Cost per task, log scale Claude Fable 5.1 GPT-6 Luna
ModelEffortHarnessScoreCost per taskOutput tokens
Claude Fable 5.1 76.7% $16
Gemini 4 Carbon 76.5% $5.76
Claude Sonnet 5.5 75.9% $8.79
Claude Opus 5.5 75.8% $11
Muse Wonder Full 75% $4.58
Gemini 4 Argon 74.8% $12
GPT-6.1 Terra 73.3% $0.56
Gemini 3.8 Flash 72.9% $8.24
GPT-6 Astra 72.7% $5.83
Muse Wonder 72.5% $5.60
Qwen3.8 Max (0902) 72% $4.43
Grok 4.7 71.5% $6.48
Kimi K3 71.3% $3.07
GPT-6.1 Sol 71.1% $1.14
Kimi K3.1 Pro 70% $1.59
GPT-6 Luna 70% $0.14
Step 5 Preview 69.5% $1.33
DeepSeek V4.1 Flash 1017 69.2% $0.33
GLM-5.3 69.1% $3.79
Mistral Large 4 69% $2.47
Tencent Hy4 preview 69% $0.79
MiMo-V2.6 Flash 68.9% $0.11
Mistral Large 4 (Le Chonk, open weights) 68.7% $0.83
MiniMax-M3.1-Flash 67.8% $0.27
DeepSeek V4.1 Flash 67.8% $0.23
MiMo-V2.6 Pro 67.3% $0.33
GLM-5.3-Flash 66.3% $0.71
Qwen 3.8 Max 66% $2.66
Claude Haiku 5.5 65.5% $1.36
DeepSeek V4 Pro 0813 65.1% $1.08
Beam 1.0 64.7%
Muse Spark Open 62% $5.35
MiniMax-M3 59.8% $3.18
Mercury 3 57.9% $0.19
Nemotron 3.5 Super VL 52.8% $1.29
Inkling 41% $1.67
Inkling-Small 40.8% $0.25

Number one over time

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

Anthropic
Own leaderboard 1 leader
Claude Fable 5.1 76.7%

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 2027 Near the ceiling (85%) Claude Fable 5.1 Kimi K3 MiniMax-M3 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Fable 5.1 MiniMax-M3

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.

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.

Anthropic
Excel Modeling Benchmark v1.1 3 leaders
Claude Fable 5.1 76.7%
Excel Modeling Benchmark 2 leaders
Claude Fable 5.1 76.7%
Excel Modeling Benchmark v1.1 1 leader
Claude Fable 5.1 76.7%
Replaces Excel Modeling 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%