Benchmark Almanac

Insurance Agent Benchmark

Cooper + Gemini 3.8 Flash 85.6% tops its leaderboard Grok 4.6 77.6% leads as a model on its own
Not in model cards

Models read commercial insurance documents and answer questions about their contents. Cooper compares each model alone and with its document processing system.

Phase one compares 17 models on 166 documents across 11 test tracks. A model judge scores answers against human-verified references.

Made by
Cooper Labs
Released
September 2026
Kind
Documents, charts and text in images
Level
University
Scored
Graded by a model
Answers
Open-ended
Tasks
Kept private
Measures
Capability
Language
English
In and out
Text, Documents → Text
Interaction
Single turn
Items from
Real work and use
Refresh
One fixed set
Run by
A host
In views
Law and finance
Strongest model tested
Claude Fable 5.1 #8 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
34
Entries on its own leaderboard, 17 models we track

Cooper + Gemini 3.8 Flash tops the leaderboard with 85.6% Grok 4.6 leads with 77.6% among models released by September 2026

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 on its own is Grok 4.6, at 77.6%. Agents built around a model, such as Cooper + Gemini 3.8 Flash, 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.

The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Cooper + Gemini 3.8 Flash 85.6%
  2. 2 Cooper + Claude Opus 5 85.3%
  3. 3 Cooper + Claude Fable 5 84.6%
  4. 4 Cooper + Claude Sonnet 5 84.4%
  5. 5 Cooper + Gemini 3.6 Flash 83.3%
  6. 6 Cooper + Gemini 3.7 Flash 83.3%
  7. 7 Cooper + Claude Fable 5.1 82.3%
  8. 8 Cooper + Muse Spark 1.3 82%
  9. 9 Cooper + Claude Opus 4.6 81.5%
  10. 10 Cooper + Claude Sonnet 4.6 81.5%
  11. 11 Cooper + GPT-5.6 Terra 80.6%
  12. 12 Cooper + GPT-5.6 Sol 80.5%
  13. 13 Cooper + Gemini 3.1 Pro 79.6%
  14. 14 Cooper + Grok 4.6 78.7%
  15. 15 Cooper + GPT-5.6 Luna 77.7%
  16. 16 Grok 4.6 Model alone 77.6%
  17. 17 Cooper + Gemini 3.5 Flash-Lite 77.2%
  18. 18 GPT-5.6 Terra Model alone 76.8%
  19. 19 GPT-5.6 Sol Model alone 76.7%
  20. 20 Gemini 3.7 Flash Model alone 76.7%
  21. 21 GPT-5.6 Luna Model alone 76.5%
  22. 22 Gemini 3.8 Flash Model alone 75.9%
  23. 23 Gemini 3.6 Flash Model alone 74.9%
  24. 24 Cooper + Claude Haiku 4.5 74.8%
  25. 25 Gemini 3.1 Pro Model alone 74.6%
  26. 26 Claude Opus 5 Model alone 73.4%
  27. 27 Gemini 3.5 Flash-Lite Model alone 73.2%
  28. 28 Claude Fable 5.1 Model alone 72.9%
  29. 29 Claude Sonnet 5 Model alone 71.7%
  30. 30 Claude Fable 5 Model alone 70.7%
  31. 31 Claude Opus 4.6 Model alone 70.6%
  32. 32 Claude Sonnet 4.6 Model alone 69.6%
  33. 33 Muse Spark 1.3 Model alone 66.9%
  34. 34 Claude Haiku 4.5 Model alone 59.1%

Source: Official Insurance Agent Benchmark leaderboard

#EntryScoreSettingDateSource
1 Cooper + Gemini 3.8 Flash 85.6% Sep 2026 Official leaderboard
2 Cooper + Claude Opus 5 85.3% Jul 2026 Official leaderboard
3 Cooper + Claude Fable 5 84.6% Jun 2026 Official leaderboard
4 Cooper + Claude Sonnet 5 84.4% Jun 2026 Official leaderboard
5 Cooper + Gemini 3.6 Flash 83.3% Jul 2026 Official leaderboard
6 Cooper + Gemini 3.7 Flash 83.3% Aug 2026 Official leaderboard
7 Cooper + Claude Fable 5.1 82.3% Sep 2026 Official leaderboard
8 Cooper + Muse Spark 1.3 82% Sep 2026 Official leaderboard
9 Cooper + Claude Opus 4.6 81.5% Feb 2026 Official leaderboard
10 Cooper + Claude Sonnet 4.6 81.5% Feb 2026 Official leaderboard
11 Cooper + GPT-5.6 Terra 80.6% Jul 2026 Official leaderboard
12 Cooper + GPT-5.6 Sol 80.5% Jul 2026 Official leaderboard
13 Cooper + Gemini 3.1 Pro 79.6% Feb 2026 Official leaderboard
14 Cooper + Grok 4.6 78.7% Aug 2026 Official leaderboard
15 Cooper + GPT-5.6 Luna 77.7% Jul 2026 Official leaderboard
16 Grok 4.6 77.6% Model alone Aug 2026 Official leaderboard
17 Cooper + Gemini 3.5 Flash-Lite 77.2% Jul 2026 Official leaderboard
18 GPT-5.6 Terra 76.8% Model alone Jul 2026 Official leaderboard
19 GPT-5.6 Sol 76.7% Model alone Jul 2026 Official leaderboard
20 Gemini 3.7 Flash 76.7% Model alone Aug 2026 Official leaderboard
21 GPT-5.6 Luna 76.5% Model alone Jul 2026 Official leaderboard
22 Gemini 3.8 Flash 75.9% Model alone Sep 2026 Official leaderboard
23 Gemini 3.6 Flash 74.9% Model alone Jul 2026 Official leaderboard
24 Cooper + Claude Haiku 4.5 74.8% Oct 2025 Official leaderboard
25 Gemini 3.1 Pro 74.6% Model alone Feb 2026 Official leaderboard
26 Claude Opus 5 73.4% Model alone Jul 2026 Official leaderboard
27 Gemini 3.5 Flash-Lite 73.2% Model alone Jul 2026 Official leaderboard
28 Claude Fable 5.1 72.9% Model alone Sep 2026 Official leaderboard
29 Claude Sonnet 5 71.7% Model alone Jun 2026 Official leaderboard
30 Claude Fable 5 70.7% Model alone Jun 2026 Official leaderboard
31 Claude Opus 4.6 70.6% Model alone Feb 2026 Official leaderboard
32 Claude Sonnet 4.6 69.6% Model alone Feb 2026 Official leaderboard
33 Muse Spark 1.3 66.9% Model alone Sep 2026 Official leaderboard
34 Claude Haiku 4.5 59.1% Model alone Oct 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Grok 4.6 77.6% Model alone Aug 2026 Official leaderboard
2 GPT-5.6 Terra 76.8% Model alone Jul 2026 Official leaderboard
3 GPT-5.6 Sol 76.7% Model alone Jul 2026 Official leaderboard
4 Gemini 3.7 Flash 76.7% Model alone Aug 2026 Official leaderboard
5 GPT-5.6 Luna 76.5% Model alone Jul 2026 Official leaderboard
6 Gemini 3.8 Flash 75.9% Model alone Sep 2026 Official leaderboard
7 Gemini 3.6 Flash 74.9% Model alone Jul 2026 Official leaderboard
8 Gemini 3.1 Pro 74.6% Model alone Feb 2026 Official leaderboard
9 Claude Opus 5 73.4% Model alone Jul 2026 Official leaderboard
10 Gemini 3.5 Flash-Lite 73.2% Model alone Jul 2026 Official leaderboard
11 Claude Fable 5.1 72.9% Model alone Sep 2026 Official leaderboard
12 Claude Sonnet 5 71.7% Model alone Jun 2026 Official leaderboard
13 Claude Fable 5 70.7% Model alone Jun 2026 Official leaderboard
14 Claude Opus 4.6 70.6% Model alone Feb 2026 Official leaderboard
15 Claude Sonnet 4.6 69.6% Model alone Feb 2026 Official leaderboard
16 Muse Spark 1.3 66.9% Model alone Sep 2026 Official leaderboard
17 Claude Haiku 4.5 59.1% Model alone Oct 2025 Official leaderboard

The best score, 85.6%, costs $0.41 a task

Runs on the cost frontier score higher than any run that's cheaper. Cooper + Gemini 3.7 Flash comes within 2.3 points of the best for $0.20 a task, about half the cost. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

xAI Google DeepMind 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% $0.01 $0.10 $1.00 $10 Cost per task, log scale Cooper + Gemini 3.8 Flash Cooper + GPT-5.6 Luna Gemini 3.7 Flash, Model alone 0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 $10 Cost per task, log scale Cooper + Gemini 3.8 Flash Cooper + GPT-5.6 Luna

Source: Official Insurance Agent Benchmark leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Cooper + Gemini 3.8 Flash Cooper 85.6% $0.41
Cooper + Claude Opus 5 Cooper 85.3% $3.37
Cooper + Claude Fable 5 Cooper 84.6% $6.63
Cooper + Claude Sonnet 5 Cooper 84.4% $1.34
Cooper + Gemini 3.6 Flash Cooper 83.3% $0.42
Cooper + Gemini 3.7 Flash Cooper 83.3% $0.20
Cooper + Claude Fable 5.1 Cooper 82.3% $5.34
Cooper + Muse Spark 1.3 Cooper 82% $0.56
Cooper + Claude Opus 4.6 Cooper 81.5% $2.79
Cooper + Claude Sonnet 4.6 Cooper 81.5% $1.62
Cooper + GPT-5.6 Terra Cooper 80.6% $1.08
Cooper + GPT-5.6 Sol Cooper 80.5% $1.34
Cooper + Gemini 3.1 Pro Cooper 79.6% $1.17
Cooper + Grok 4.6 Cooper 78.7% $0.21
Cooper + GPT-5.6 Luna Cooper 77.7% $0.11
Grok 4.6 Model alone 77.6% $0.07
Cooper + Gemini 3.5 Flash-Lite Cooper 77.2% $0.20
GPT-5.6 Terra Model alone 76.8% $0.10
GPT-5.6 Sol Model alone 76.7% $0.13
Gemini 3.7 Flash Model alone 76.7% $0.01
GPT-5.6 Luna Model alone 76.5% $0.01
Gemini 3.8 Flash Model alone 75.9% $0.01
Gemini 3.6 Flash Model alone 74.9% $0.01
Cooper + Claude Haiku 4.5 Cooper 74.8% $0.55
Gemini 3.1 Pro Model alone 74.6% $0.03
Claude Opus 5 Model alone 73.4% $0.25
Gemini 3.5 Flash-Lite Model alone 73.2% $0.01
Claude Fable 5.1 Model alone 72.9% $0.53
Claude Sonnet 5 Model alone 71.7% $0.11
Claude Fable 5 Model alone 70.7% $0.38
Claude Opus 4.6 Model alone 70.6% $0.23
Claude Sonnet 4.6 Model alone 69.6% $0.14
Muse Spark 1.3 Model alone 66.9% $0.04
Claude Haiku 4.5 Model alone 59.1% $0.01

The best score, 77.6%, costs $0.07 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-5.6 Luna at Model alone comes within 1.1 points of the best for $0.01 a task, 7 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

xAI Google DeepMind Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.01 $0.02 $0.05 $0.10 $0.20 $0.50 $1.00 Cost per task, log scale Grok 4.6, Model alone Gemini 3.7 Flash, Model alone 0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 Cost per task, log scale Grok 4.6, Model alone

Source: Official Insurance Agent Benchmark leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Grok 4.6 Model alone 77.6% $0.07
GPT-5.6 Terra Model alone 76.8% $0.10
GPT-5.6 Sol Model alone 76.7% $0.13
Gemini 3.7 Flash Model alone 76.7% $0.01
GPT-5.6 Luna Model alone 76.5% $0.01
Gemini 3.8 Flash Model alone 75.9% $0.01
Gemini 3.6 Flash Model alone 74.9% $0.01
Gemini 3.1 Pro Model alone 74.6% $0.03
Claude Opus 5 Model alone 73.4% $0.25
Gemini 3.5 Flash-Lite Model alone 73.2% $0.01
Claude Fable 5.1 Model alone 72.9% $0.53
Claude Sonnet 5 Model alone 71.7% $0.11
Claude Fable 5 Model alone 70.7% $0.38
Claude Opus 4.6 Model alone 70.6% $0.23
Claude Sonnet 4.6 Model alone 69.6% $0.14
Muse Spark 1.3 Model alone 66.9% $0.04
Claude Haiku 4.5 Model alone 59.1% $0.01

Number one over time

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

xAI
Own leaderboard Not tested since Sep 2026
Grok 4.6 77.6%

Source: Official Insurance Agent Benchmark leaderboard

Scores over time

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% 2026 Near the ceiling (85%) Grok 4.6 Gemini 3.1 Pro Claude Haiku 4.5 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Grok 4.6 Claude Opus 4.6

Source: Official Insurance Agent Benchmark 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.

More in Vision

All 636 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. Qwen 3.6 Plus 86%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy. Seed2.1 Pro 90.7%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (34 entries).
  • Pass Models scored 17 models we track, plus 17 other entries on its leaderboard.
  • Pass Newest score The newest is from September 2026, for Gemini 3.8 Flash.
  • Note Frontier models Only 1 of the top 10 has a score. The strongest tested is Claude Fable 5.1 (#8).
  • Pass Headroom The best is 77.6%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on askcooper.ai: 34 rows for models we track 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