AA-AnalystAgent - Pass@K: leaderboard
Artificial Analysis evaluates agents on data-analysis work. The published headline score, pass-at-one and pass-at-k rates are retained as separate measurements of task success.
Metric: Pass@K (%). Source: artificialanalysis.ai. Saturation forecast: Not forecast. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Luna (Max) | 85 | #61 (GPT-6 Luna) |
| 2 | Claude Haiku 5.5 (Max) | 83.75 | #43 (Claude Haiku 5.5) |
| 3 | Gemini 3.1 Pro (Preview) | 81.25 | #64 |
| 4 | DeepSeek V4.1 Flash (Max) | 80 | #37 (DeepSeek V4.1 Flash) |
| 5 | Inkling Small | 80 | #188 |
| 6 | MiMo-V2.6-Pro | 80 | #48 |
| 7 | GLM-5.3 Flash | 78.75 | #77 |
| 8 | GPT-5.5 (xHigh) | 78.75 | #33 (GPT-5.5) |
| 9 | Claude Opus 4.8 (Max) | 78.75 | #39 (Claude Opus 4.8) |
| 10 | Inkling (xHigh) | 78.75 | #144 (Inkling) |
| 11 | Gemini 3.7 Flash (High) | 77.5 | #24 (Gemini 3.7 Flash) |
| 12 | Claude Sonnet 5 (Max) | 77.5 | #65 (Claude Sonnet 5) |
| 13 | Gemini 3.5 Flash (High) | 76.25 | #66 (Gemini 3.5 Flash) |
| 14 | Grok 4.5 (High) | 76.25 | #54 (Grok 4.5) |
| 15 | Grok 4.6 (High) | 76.25 | #29 (Grok 4.6) |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aa-analystagent-pass-k · How It Works · Data refreshed daily, snapshot 2026-11-05.