CorporateBench: leaderboard
An agent extracts and queries facts across corporate document collections of a few hundred to more than 230,000 emails and files; the score is the mean over 3 Fireworks runs of all 128 tasks. It tests retrieval at e-discovery scale rather than drafting legal work product.
Metric: Mean score over 3 runs of 128 tasks (%). Source: fireworks.ai. Saturation forecast: Around July 2027. 12 models tracked. Updated 16 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | GPT-6.1 Sol | Max | 82.21 | #10 |
| 2 | GPT-6 Astra | Max | 81.96 | #6 |
| 3 | GPT-6 Sol | Max | 79 | #26 |
| 4 | Kimi K3.1 Pro | Max | 78.45 | #29 |
| 5 | Gemini 3.8 Flash | High | 77.57 | #36 |
| 6 | Claude Opus 5.5 | Max | 75.38 | #5 |
| 7 | DeepSeek V4.1 Flash | Max | 75.29 | #62 |
| 8 | DeepSeek V4.1 Flash (1017) | Max | 75.21 | #41 |
| 9 | Claude Opus 5 | Max | 74 | #20 |
| 10 | Muse Spark Open | Max | 73 | #81 |
| 11 | Kimi K3 | Max | 72.33 | #56 |
| 12 | GPT-6 Luna | Max | 71.74 | #99 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=corporatebench · How It Works · Data refreshed daily, snapshot 2026-11-16.