Insurance Agent Benchmark (Model Alone): leaderboard
Insurance Agent Benchmark (Cooper Labs), phase one: 166 real commercial-lines documents — ACORDs, loss runs, statements of value, policy wordings, submission emails with attachments, scanned and corrupt files — asked one question each and graded by an LLM judge against a human-verified answer key, where an absent or unreadable value must be reported as absent. This board is the raw single call: the file and the question go straight to the model with no scaffold.
Metric: Judge Accuracy (%). Source: www.askcooper.ai. Saturation forecast: Not forecast. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.6 | 77.6 | #29 |
| 2 | GPT-5.6 Terra | 76.8 | #40 |
| 3 | GPT-5.6 Sol | 76.7 | #22 |
| 4 | Gemini 3.7 Flash | 76.7 | #24 |
| 5 | GPT-5.6 Luna | 76.5 | #79 |
| 6 | Gemini 3.8 Flash | 75.9 | #30 |
| 7 | Gemini 3.6 Flash | 74.9 | #57 |
| 8 | Gemini 3.1 Pro (Preview) | 74.6 | #64 |
| 9 | Claude Opus 5 | 73.4 | #16 |
| 10 | Gemini 3.5 Flash Lite | 73.2 | #174 |
| 11 | Claude Fable 5.1 | 72.9 | #10 |
| 12 | Claude Sonnet 5 | 71.7 | #65 |
| 13 | Claude Fable 5 | 70.7 | #20 |
| 14 | Claude Opus 4.6 | 70.6 | #67 |
| 15 | Claude Sonnet 4.6 | 69.6 | #94 |
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-model-alone · How It Works · Data refreshed daily, snapshot 2026-11-05.