Benchmark Almanac

LevelBench-100 v5.0

Llama 4 Maverick 17B-128E 100% tops its leaderboard Gemini 4 Carbon 95.2% leads the models we track
Not in model cards
Version 5.0

Models use ledger exports and supporting workbooks to close small business accounts. They reconcile transactions, apply accounting rules, and identify missing records.

Level ranks 21 distinct unassisted models across three overlapping test groups. The 68 rows include reasoning settings and runs with playbooks and operator review.

Made by
Level
Released
November 2026
Kind
Finance
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
Public, plus a private test set
Measures
Capability
Language
English
In and out
Text, Tables → Tables, Text, Actions
Interaction
Agentic
Environment
Desktop apps
Items from
Real work and use
Run by
A host
Strongest model tested
Claude Fable 5.2 #2 on the capability index
Last model added
November 2026
Scores refreshed
November 20th, 2026
42
Entries on its own leaderboard, 19 models we track

Llama 4 Maverick 17B-128E tops the leaderboard with 100% Gemini 4 Carbon leads with 95.2%

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran on its own with several settings, we keep its best. Every run is in the next chart. Agents other companies built around a model get a pale shade of that model's lab color. Gray marks blends and entries not in our model list, such as fine-tunes and older models. None of these count toward leaders. The best model we track is Gemini 4 Carbon, at 95.2%. Agents built around a model, such as Level Playbooks + DeepSeek-R1, are their own entries. We don't count them as the model's score. This view shows only models we track, each at its best.

  1. 1 Llama 4 Maverick 17B-128E 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  2. 2 Moonshot Kimi K3.1 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  3. 3 Level Playbooks + DeepSeek-R1 May 2025, 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Reasoning 100%
  4. 4 Level Playbooks + Qwen3-Next-80B-A3B 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / A3B 100%
  5. 5 Level Playbooks + Gemini 2.5 Flash-Lite Sep 2025, 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  6. 6 Level Playbooks + Gemini 2.5 Flash Sep 2025, 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  7. 7 Level Playbooks + DeepSeek-V3.2 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  8. 8 Level Playbooks + Mistral Large 3 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / 675B 100%
  9. 9 Level Playbooks + Gemini 3 Flash 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low 100%
  10. 10 Level Playbooks + Kimi K2.5 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  11. 11 Level Playbooks + Gemini 3.5 Flash 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  12. 12 Level Playbooks + Gemini 3.8 Flash 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  13. 13 Level Playbooks + GPT-6 Astra 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low 100%
  14. 14 Level Playbooks + Claude Sonnet 5.5 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  15. 15 Level Playbooks + GPT-6.1 Sol 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium 100%
  16. 16 Level Playbooks + Gemini 4 Argon 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium 100%
  17. 17 Level Playbooks + Claude Haiku 5.5 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  18. 18 Level Playbooks + GPT-6.1 Terra 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low 100%
  19. 19 Level Playbooks + Gemini 4 Carbon 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium 100%
  20. 20 Level Playbooks + Claude Fable 5.2 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) 100%
  21. 21 Level Playbooks + Gemini 4 Hydrogen 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low 100%
  22. 22 Gemini 4 Carbon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 95.2%
  23. 23 Gemini 4 Argon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 94.7%
  24. 24 GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 93.3%
  25. 25 GPT-6 Astra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 92.9%
  26. 26 Gemini 4 Hydrogen 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.7%
  27. 27 Gemini 3.8 Flash 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model 87.6%
  28. 28 GPT-6.1 Terra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 87.5%
  29. 29 Claude Fable 5.2 25-Company Sample · 372 Checks / Unassisted Raw AI Model 86.8%
  30. 30 Gemini 3.5 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model 86.3%
  31. 31 Kimi K2.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 77.2%
  32. 32 Gemini 3 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 76.2%
  33. 33 Gemini 2.5 Flash Sep 2025, 60 Client Closes · 924 Checks / Unassisted Raw AI Model 75.9%
  34. 34 Claude Sonnet 5.5 25-Company Sample · 372 Checks / Unassisted Raw AI Model 73.4%
  35. 35 Qwen3-Next-80B-A3B 60 Client Closes · 924 Checks / Unassisted Raw AI Model / A3B 71.1%
  36. 36 DeepSeek-V3.2 60 Client Closes · 924 Checks / Unassisted Raw AI Model 69.8%
  37. 37 DeepSeek-R1 May 2025, 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Reasoning 69.7%
  38. 38 Claude Haiku 5.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 66%
  39. 39 Mistral Large 3 60 Client Closes · 924 Checks / Unassisted Raw AI Model / 675B 58.1%
  40. 40 Gemini 2.5 Flash-Lite Sep 2025, 60 Client Closes · 924 Checks / Unassisted Raw AI Model 52.4%
  41. 41 Gemini 4 Argon (High) [Cut Off at 8K Limit] 25-Company Sample · 372 Checks / Raw AI Model (8K Output Cap Truncated) / High 43.8%
  42. 42 Gemini 4 Carbon (High) [Cut Off at 8K Limit] 60 Client Closes · 924 Checks / Raw AI Model (8K Output Cap Truncated) / High 0%

Source: Official LevelBench-100 v5.0 leaderboard

  1. 1 Gemini 4 Carbon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 95.2%
  2. 2 Gemini 4 Argon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 94.7%
  3. 3 GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 93.3%
  4. 4 GPT-6 Astra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 92.9%
  5. 5 Gemini 4 Hydrogen 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.7%
  6. 6 Gemini 3.8 Flash 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model 87.6%
  7. 7 GPT-6.1 Terra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 87.5%
  8. 8 Claude Fable 5.2 25-Company Sample · 372 Checks / Unassisted Raw AI Model 86.8%
  9. 9 Gemini 3.5 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model 86.3%
  10. 10 Kimi K2.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 77.2%
  11. 11 Gemini 3 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 76.2%
  12. 12 Gemini 2.5 Flash Sep 2025, 60 Client Closes · 924 Checks / Unassisted Raw AI Model 75.9%
  13. 13 Claude Sonnet 5.5 25-Company Sample · 372 Checks / Unassisted Raw AI Model 73.4%
  14. 14 Qwen3-Next-80B-A3B 60 Client Closes · 924 Checks / Unassisted Raw AI Model / A3B 71.1%
  15. 15 DeepSeek-V3.2 60 Client Closes · 924 Checks / Unassisted Raw AI Model 69.8%
  16. 16 DeepSeek-R1 May 2025, 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Reasoning 69.7%
  17. 17 Claude Haiku 5.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 66%
  18. 18 Mistral Large 3 60 Client Closes · 924 Checks / Unassisted Raw AI Model / 675B 58.1%
  19. 19 Gemini 2.5 Flash-Lite Sep 2025, 60 Client Closes · 924 Checks / Unassisted Raw AI Model 52.4%

Source: Official LevelBench-100 v5.0 leaderboard

#EntryScoreSettingDateSource
1 Llama 4 Maverick 17B-128E 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences), Level Playbooks Official leaderboard
2 Moonshot Kimi K3.1 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences), Level Playbooks Official leaderboard
3 Level Playbooks + DeepSeek-R1 May 2025 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Reasoning May 2025 Official leaderboard
4 Level Playbooks + Qwen3-Next-80B-A3B 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / A3B Sep 2025 Official leaderboard
5 Level Playbooks + Gemini 2.5 Flash-Lite Sep 2025 100% 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) Sep 2025 Official leaderboard
6 Level Playbooks + Gemini 2.5 Flash Sep 2025 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Sep 2025 Official leaderboard
7 Level Playbooks + DeepSeek-V3.2 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Dec 2025 Official leaderboard
8 Level Playbooks + Mistral Large 3 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / 675B Dec 2025 Official leaderboard
9 Level Playbooks + Gemini 3 Flash 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Dec 2025 Official leaderboard
10 Level Playbooks + Kimi K2.5 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Jan 2026 Official leaderboard
11 Level Playbooks + Gemini 3.5 Flash 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) May 2026 Official leaderboard
12 Level Playbooks + Gemini 3.8 Flash 100% 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) Sep 2026 Official leaderboard
13 Level Playbooks + GPT-6 Astra 100% 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Sep 2026 Official leaderboard
14 Level Playbooks + Claude Sonnet 5.5 100% 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) Sep 2026 Official leaderboard
15 Level Playbooks + GPT-6.1 Sol 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Sep 2026 Official leaderboard
16 Level Playbooks + Gemini 4 Argon 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Sep 2026 Official leaderboard
17 Level Playbooks + Claude Haiku 5.5 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Oct 2026 Official leaderboard
18 Level Playbooks + GPT-6.1 Terra 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Oct 2026 Official leaderboard
19 Level Playbooks + Gemini 4 Carbon 100% 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Oct 2026 Official leaderboard
20 Level Playbooks + Claude Fable 5.2 100% 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) Nov 2026 Official leaderboard
21 Level Playbooks + Gemini 4 Hydrogen 100% 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Nov 2026 Official leaderboard
22 Gemini 4 Carbon 95.2% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low Oct 2026 Official leaderboard
23 Gemini 4 Argon 94.7% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low Sep 2026 Official leaderboard
24 GPT-6.1 Sol 93.3% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High Sep 2026 Official leaderboard
25 GPT-6 Astra 92.9% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium Sep 2026 Official leaderboard
26 Gemini 4 Hydrogen 91.7% 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium Nov 2026 Official leaderboard
27 Gemini 3.8 Flash 87.6% 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model Sep 2026 Official leaderboard
28 GPT-6.1 Terra 87.5% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium Oct 2026 Official leaderboard
29 Claude Fable 5.2 86.8% 25-Company Sample · 372 Checks / Unassisted Raw AI Model Nov 2026 Official leaderboard
30 Gemini 3.5 Flash 86.3% 60 Client Closes · 924 Checks / Unassisted Raw AI Model May 2026 Official leaderboard
31 Kimi K2.5 77.2% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Jan 2026 Official leaderboard
32 Gemini 3 Flash 76.2% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High Dec 2025 Official leaderboard
33 Gemini 2.5 Flash Sep 2025 75.9% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Sep 2025 Official leaderboard
34 Claude Sonnet 5.5 73.4% 25-Company Sample · 372 Checks / Unassisted Raw AI Model Sep 2026 Official leaderboard
35 Qwen3-Next-80B-A3B 71.1% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / A3B Sep 2025 Official leaderboard
36 DeepSeek-V3.2 69.8% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Dec 2025 Official leaderboard
37 DeepSeek-R1 May 2025 69.7% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Reasoning May 2025 Official leaderboard
38 Claude Haiku 5.5 66% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Oct 2026 Official leaderboard
39 Mistral Large 3 58.1% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / 675B Dec 2025 Official leaderboard
40 Gemini 2.5 Flash-Lite Sep 2025 52.4% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Sep 2025 Official leaderboard
41 Gemini 4 Argon (High) [Cut Off at 8K Limit] 43.8% 25-Company Sample · 372 Checks / Raw AI Model (8K Output Cap Truncated) / High Official leaderboard
42 Gemini 4 Carbon (High) [Cut Off at 8K Limit] 0% 60 Client Closes · 924 Checks / Raw AI Model (8K Output Cap Truncated) / High Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 4 Carbon 95.2% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low Oct 2026 Official leaderboard
2 Gemini 4 Argon 94.7% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low Sep 2026 Official leaderboard
3 GPT-6.1 Sol 93.3% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High Sep 2026 Official leaderboard
4 GPT-6 Astra 92.9% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium Sep 2026 Official leaderboard
5 Gemini 4 Hydrogen 91.7% 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium Nov 2026 Official leaderboard
6 Gemini 3.8 Flash 87.6% 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model Sep 2026 Official leaderboard
7 GPT-6.1 Terra 87.5% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium Oct 2026 Official leaderboard
8 Claude Fable 5.2 86.8% 25-Company Sample · 372 Checks / Unassisted Raw AI Model Nov 2026 Official leaderboard
9 Gemini 3.5 Flash 86.3% 60 Client Closes · 924 Checks / Unassisted Raw AI Model May 2026 Official leaderboard
10 Kimi K2.5 77.2% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Jan 2026 Official leaderboard
11 Gemini 3 Flash 76.2% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High Dec 2025 Official leaderboard
12 Gemini 2.5 Flash Sep 2025 75.9% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Sep 2025 Official leaderboard
13 Claude Sonnet 5.5 73.4% 25-Company Sample · 372 Checks / Unassisted Raw AI Model Sep 2026 Official leaderboard
14 Qwen3-Next-80B-A3B 71.1% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / A3B Sep 2025 Official leaderboard
15 DeepSeek-V3.2 69.8% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Dec 2025 Official leaderboard
16 DeepSeek-R1 May 2025 69.7% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Reasoning May 2025 Official leaderboard
17 Claude Haiku 5.5 66% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Oct 2026 Official leaderboard
18 Mistral Large 3 58.1% 60 Client Closes · 924 Checks / Unassisted Raw AI Model / 675B Dec 2025 Official leaderboard
19 Gemini 2.5 Flash-Lite Sep 2025 52.4% 60 Client Closes · 924 Checks / Unassisted Raw AI Model Sep 2025 Official leaderboard

Rankings by part

We read 4 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Bookkeeping, Accrual Close and CFO Analysis. Gemini 4 Argon leads 3 of them, and GPT-6 Astra leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

Bookkeeping

19 models we track have a score here.

  1. 1 Gemini 4 Argon ↑1 99.8%
  2. 2 Gemini 4 Carbon ↓1 99.8%
  3. 3 Gemini 4 Hydrogen ↑2 99.7%
  4. 4 GPT-6 Astra 99.7%
  5. 5 GPT-6.1 Sol ↓2 99.7%
  6. 6 Gemini 3.8 Flash 99.5%
  7. 7 GPT-6.1 Terra 99.3%
  8. 8 Qwen3-Next-80B-A3B ↑6 99.2%
  9. 9 Claude Fable 5.2 ↓1 99.2%
  10. 10 Gemini 3.5 Flash ↓1 99.1%

The best score, 100%, costs $0.004 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.

Google DeepMind DeepSeek OpenAI Agents other companies built, in a pale shade of the model’s lab Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Level Playbooks + Gemini 2.5 Flash-Lite, 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) GPT-6.1 Terra, 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low Gemini 2.5 Flash, 60 Client Closes · 924 Checks / Unassisted Raw AI Model DeepSeek-V3.2, 60 Client Closes · 924 Checks / Unassisted Raw AI Model Gemini 2.5 Flash-Lite, 60 Client Closes · 924 Checks / Unassisted Raw AI Model 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Level Playbooks + Gemini 2.5 Flash-Lite, 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) GPT-6.1 Terra, 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low

Source: Official LevelBench-100 v5.0 leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Level Playbooks + DeepSeek-R1 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Reasoning Level Playbooks 100% $0.03
Level Playbooks + Qwen3-Next-80B-A3B 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / A3B Level Playbooks 100% $0.02
Level Playbooks + Gemini 2.5 Flash-Lite 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.004
Level Playbooks + Gemini 2.5 Flash-Lite 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.004
Level Playbooks + Gemini 2.5 Flash 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.008
Level Playbooks + DeepSeek-V3.2 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.01
Level Playbooks + Mistral Large 3 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / 675B Level Playbooks 100% $0.04
Level Playbooks + Gemini 3 Flash 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Level Playbooks 100% $0.02
Level Playbooks + Gemini 3 Flash 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / High Level Playbooks 100% $0.02
Level Playbooks + Kimi K2.5 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.02
Level Playbooks + Gemini 3.5 Flash 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.008
Level Playbooks + Gemini 3.8 Flash 374 Client Closes · 5,923 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.02
Level Playbooks + Gemini 3.8 Flash 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.02
Level Playbooks + GPT-6 Astra 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Level Playbooks 100% $0.41
Level Playbooks + GPT-6 Astra 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Level Playbooks 100% $0.42
Level Playbooks + Claude Sonnet 5.5 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.17
Level Playbooks + GPT-6.1 Sol 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Level Playbooks 100% $0.10
Level Playbooks + Gemini 4 Argon 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Level Playbooks 100% $0.11
Level Playbooks + Claude Haiku 5.5 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.07
Level Playbooks + GPT-6.1 Terra 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Level Playbooks 100% $0.02
Level Playbooks + GPT-6.1 Terra 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Level Playbooks 100% $0.02
Level Playbooks + Gemini 4 Carbon 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Level Playbooks 100% $0.11
Level Playbooks + Gemini 4 Carbon 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Level Playbooks 100% $0.10
Level Playbooks + Gemini 4 Carbon 60 Client Closes · 924 Checks / AI + Level Playbooks (2,200+ Past Experiences) / High Level Playbooks 100% $0.12
Level Playbooks + Claude Fable 5.2 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) Level Playbooks 100% $0.63
Level Playbooks + Gemini 4 Hydrogen 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Low Level Playbooks 100% $0.08
Level Playbooks + Gemini 4 Hydrogen 25-Company Sample · 372 Checks / AI + Level Playbooks (2,200+ Past Experiences) / Medium Level Playbooks 100% $0.08
Gemini 4 Carbon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 95.2% $0.06
Gemini 4 Carbon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 95.1% $0.10
Gemini 4 Argon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 94.7% $0.05
Gemini 4 Argon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 94.5% $0.05
Gemini 4 Carbon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 93.5% $0.07
GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 93.3% $0.07
GPT-6 Astra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 92.9% $0.02
GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 92% $0.04
Gemini 4 Argon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.9% $0.06
GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 91.7% $0.03
Gemini 4 Carbon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.7% $0.09
Gemini 4 Hydrogen 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.7% $0.08
Gemini 4 Argon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 91.1% $0.06
Gemini 3.8 Flash 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model 87.6% $0.02
GPT-6.1 Terra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 87.5% $0.006
GPT-6 Astra 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 87.4% $0.15
GPT-6 Astra 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 87.1% $0.20
Claude Fable 5.2 25-Company Sample · 372 Checks / Unassisted Raw AI Model 86.8% $0.34
Gemini 3.5 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model 86.3% $0.009
Gemini 4 Hydrogen 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 85.8% $0.04
GPT-6.1 Terra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 85.6% $0.004
Gemini 3.8 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model 79.9% $0.02
Kimi K2.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 77.2% $0.007
Gemini 3 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 76.2% $0.01
Gemini 2.5 Flash Sep 2025 60 Client Closes · 924 Checks / Unassisted Raw AI Model 75.9% $0.003
Claude Sonnet 5.5 25-Company Sample · 372 Checks / Unassisted Raw AI Model 73.4% $0.13
Gemini 3 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 72.6% $0.005
Qwen3-Next-80B-A3B 60 Client Closes · 924 Checks / Unassisted Raw AI Model / A3B 71.1% $0.008
DeepSeek-V3.2 60 Client Closes · 924 Checks / Unassisted Raw AI Model 69.8% $0.003
DeepSeek-R1 May 2025 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Reasoning 69.7% $0.03
Claude Haiku 5.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 66% $0.05
Mistral Large 3 60 Client Closes · 924 Checks / Unassisted Raw AI Model / 675B 58.1% $0.007
Gemini 2.5 Flash-Lite Sep 2025 60 Client Closes · 924 Checks / Unassisted Raw AI Model 52.4% $0.002
Gemini 2.5 Flash-Lite Sep 2025 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model 52.4% $0.003

The best score, 95.2%, costs $0.06 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-6 Astra at 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium comes within 2.4 points of the best for $0.02 a task, 3 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Google DeepMind DeepSeek OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Gemini 4 Carbon, 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low Gemini 3.8 Flash, 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model Gemini 2.5 Flash, 60 Client Closes · 924 Checks / Unassisted Raw AI Model DeepSeek-V3.2, 60 Client Closes · 924 Checks / Unassisted Raw AI Model Gemini 2.5 Flash-Lite, 60 Client Closes · 924 Checks / Unassisted Raw AI Model 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Gemini 4 Carbon, 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low Gemini 3.8 Flash, 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model

Source: Official LevelBench-100 v5.0 leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Carbon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 95.2% $0.06
Gemini 4 Carbon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 95.1% $0.10
Gemini 4 Argon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 94.7% $0.05
Gemini 4 Argon 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 94.5% $0.05
Gemini 4 Carbon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 93.5% $0.07
GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 93.3% $0.07
GPT-6 Astra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 92.9% $0.02
GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 92% $0.04
Gemini 4 Argon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.9% $0.06
GPT-6.1 Sol 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 91.7% $0.03
Gemini 4 Carbon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.7% $0.09
Gemini 4 Hydrogen 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 91.7% $0.08
Gemini 4 Argon 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 91.1% $0.06
Gemini 3.8 Flash 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model 87.6% $0.02
GPT-6.1 Terra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Medium 87.5% $0.006
GPT-6 Astra 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 87.4% $0.15
GPT-6 Astra 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Medium 87.1% $0.20
Claude Fable 5.2 25-Company Sample · 372 Checks / Unassisted Raw AI Model 86.8% $0.34
Gemini 3.5 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model 86.3% $0.009
Gemini 4 Hydrogen 25-Company Sample · 372 Checks / Unassisted Raw AI Model / Low 85.8% $0.04
GPT-6.1 Terra 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 85.6% $0.004
Gemini 3.8 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model 79.9% $0.02
Kimi K2.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 77.2% $0.007
Gemini 3 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model / High 76.2% $0.01
Gemini 2.5 Flash Sep 2025 60 Client Closes · 924 Checks / Unassisted Raw AI Model 75.9% $0.003
Claude Sonnet 5.5 25-Company Sample · 372 Checks / Unassisted Raw AI Model 73.4% $0.13
Gemini 3 Flash 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Low 72.6% $0.005
Qwen3-Next-80B-A3B 60 Client Closes · 924 Checks / Unassisted Raw AI Model / A3B 71.1% $0.008
DeepSeek-V3.2 60 Client Closes · 924 Checks / Unassisted Raw AI Model 69.8% $0.003
DeepSeek-R1 May 2025 60 Client Closes · 924 Checks / Unassisted Raw AI Model / Reasoning 69.7% $0.03
Claude Haiku 5.5 60 Client Closes · 924 Checks / Unassisted Raw AI Model 66% $0.05
Mistral Large 3 60 Client Closes · 924 Checks / Unassisted Raw AI Model / 675B 58.1% $0.007
Gemini 2.5 Flash-Lite Sep 2025 60 Client Closes · 924 Checks / Unassisted Raw AI Model 52.4% $0.002
Gemini 2.5 Flash-Lite Sep 2025 374 Client Closes · 5,923 Checks / Unassisted Raw AI Model 52.4% $0.003

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 Passed 85% in Nov 2026, by Gemini 4 Carbon

Source: Official LevelBench-100 v5.0 leaderboard

Gemini 3.5 Flash on the official leaderboard passed 85% in May 2026

Models released before it came out already scored that high. 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% 2025 2026 2027 Near the ceiling (85%) Gemini 4 Carbon Gemini 3.8 Flash Kimi K2.5 Gemini 2.5 Flash Qwen3-Next-80B-A3B 0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) Gemini 4 Carbon Gemini 3.8 Flash

Source: Official LevelBench-100 v5.0 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.

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.

OpenAI Google DeepMind
LevelBench-100 v5.0 Passed 85% in Nov 2026, by Gemini 4 Carbon
LevelBench-100 v5.0 Passed 85% in Nov 2026, by Gemini 4 Carbon

Source: Official leaderboards

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

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (42 entries).
  • Pass Models scored 19 models we track, plus 23 other entries on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Gemini 4 Hydrogen.
  • Pass Frontier models 7 of the top 10 have a score, led by Claude Fable 5.2 (#2).
  • Note Headroom Saturated. The best is 95.2%, past 85% since May 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on levelcfo.com: 61 rows for models we track and 4 other entries 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