Mistral Large 4 Release - FinWorkBench (Finch): leaderboard
Mistral Large 4 open-weight release post (2026-10-29): FinWorkBench (Finch) real-world finance and accounting workflows, including creating and editing spreadsheets, as reported by Mistral for five models. Harness and reasoning effort not stated. Score (%).
Metric: Score (%). Source: mistral.ai. Saturation forecast: Around February 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K3 | 77.33 | #35 |
| 2 | DeepSeek V4 Pro (0813) | 67.44 | #58 |
| 3 | Mistral Large 4 | 67.4 | #106 |
| 4 | GLM-5.3 | 65.12 | #44 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mistral-large-4-release-finworkbench-finch · How It Works · Data refreshed daily, snapshot 2026-11-05.