Harvey LAB-AA - Banking & Finance: leaderboard
Artificial Analysis's implementation of Harvey's Legal Agent Benchmark restricted to the Banking & Finance practice area (about five of its 120 private tasks): agents read case documents and draft legal deliverables graded criterion by criterion by an LLM judge. Share of rubric criteria passed, 0-100 (%); higher is better.
Metric: Criterion Pass Rate (%). Source: artificialanalysis.ai. Saturation forecast: Estimated already saturated. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Muse Spark 1.3 (Max) | 97.64 | #31 (Muse Spark 1.3) |
| 2 | Gemini 4 Argon (High) | 97.01 | #9 (Gemini 4 Argon) |
| 3 | Muse Wonder (Max) | 97.01 | #18 (Muse Wonder) |
| 4 | Muse Wonder Full (Max) | 96.64 | #12 (Muse Wonder Full) |
| 5 | Grok 4.7 (xHigh) | 95.4 | #26 (Grok 4.7) |
| 6 | Mistral Large 4 | 95.27 | #106 |
| 7 | Kimi K3 (Max) | 94.9 | #35 (Kimi K3) |
| 8 | GPT-6 Astra (Max) | 94.15 | #6 (GPT-6 Astra) |
| 9 | GPT-6.1 Sol (Max) | 93.91 | #7 (GPT-6.1 Sol) |
| 10 | GLM-5.3 (Max) | 93.66 | #44 (GLM-5.3) |
| 11 | GLM-5.3 Flash | 93.28 | #77 |
| 12 | Gemini 3.8 Flash (High) | 92.41 | #30 (Gemini 3.8 Flash) |
| 13 | DeepSeek V4.1 Flash (Max) | 92.04 | #37 (DeepSeek V4.1 Flash) |
| 14 | Claude Haiku 5.5 (High) | 91.29 | #43 (Claude Haiku 5.5) |
| 15 | Claude Haiku 5.5 (Medium) | 91.29 | #43 (Claude Haiku 5.5) |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=harvey-lab-aa-banking-finance · How It Works · Data refreshed daily, snapshot 2026-11-05.