Insurance Agent Benchmark (Cooper Harness): leaderboard
The same 166 Insurance Agent Benchmark cases run through Cooper's production document system — file-type routing, format-specific extraction or rendering, and chunked map-reduce for large files — with the candidate model doing all reading and reduction. Reported beside the model-alone run so the harness contribution stays visible.
Metric: Judge Accuracy (%). Source: www.askcooper.ai. Saturation forecast: Around February 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.8 Flash | 85.6 | #30 |
| 2 | Claude Opus 5 | 85.3 | #16 |
| 3 | Claude Fable 5 | 84.6 | #20 |
| 4 | Claude Sonnet 5 | 84.4 | #65 |
| 5 | Gemini 3.6 Flash | 83.3 | #57 |
| 6 | Gemini 3.7 Flash | 83.3 | #24 |
| 7 | Claude Fable 5.1 | 82.3 | #10 |
| 8 | Muse Spark 1.3 | 82 | #31 |
| 9 | Claude Sonnet 4.6 | 81.5 | #94 |
| 10 | Claude Opus 4.6 | 81.5 | #67 |
| 11 | GPT-5.6 Terra | 80.6 | #40 |
| 12 | GPT-5.6 Sol | 80.5 | #22 |
| 13 | Gemini 3.1 Pro (Preview) | 79.6 | #64 |
| 14 | Grok 4.6 | 78.7 | #29 |
| 15 | GPT-5.6 Luna | 77.7 | #79 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=insurance-agent-benchmark-cooper-harness · How It Works · Data refreshed daily, snapshot 2026-11-05.