Vals AI Excel Modeling v1.1: leaderboard
EMB v1.1: financial workbook creation in template and scratch modes; equal-weight seven-category accuracy. Template cells match gold workbooks after Microsoft Excel recalculation; scratch rubrics use GPT-6 Luna. Revised judge and sheet-name matching differ from v1.0; 3.5-hour limit and private scored tasks.
Metric: Accuracy (%). Source: www.vals.ai. Saturation forecast: Around November 2027. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (Max) | 76.73 | #9 (Claude Fable 5.1) |
| 2 | Claude Sonnet 5.5 (Max) | 75.9 | #11 (Claude Sonnet 5.5) |
| 3 | Claude Opus 5.5 (Max) | 75.84 | #4 (Claude Opus 5.5) |
| 4 | Muse Wonder Full (Max) | 75 | #10 (Muse Wonder Full) |
| 5 | Gemini 4 Argon (High) | 74.83 | #7 (Gemini 4 Argon) |
| 6 | Gemini 3.8 Flash (High) | 72.87 | #23 (Gemini 3.8 Flash) |
| 7 | GPT-6 Astra (Max) | 72.68 | #6 (GPT-6 Astra) |
| 8 | Muse Wonder (Max) | 72.5 | #14 (Muse Wonder) |
| 9 | Grok 4.7 (xHigh) | 71.5 | #27 (Grok 4.7) |
| 10 | Kimi K3 (Max) | 71.26 | #31 (Kimi K3) |
| 11 | GPT-6.1 Sol (Max) | 71.09 | #8 (GPT-6.1 Sol) |
| 12 | GPT-6 Luna (Max) | 69.95 | #59 (GPT-6 Luna) |
| 13 | Step 5 Preview | 69.55 | #44 |
| 14 | DeepSeek V4.1 Flash (1017) (Max) | 69.17 | #20 (DeepSeek V4.1 Flash (1017)) |
| 15 | GLM-5.3 (Max) | 69.11 | #39 (GLM-5.3) |
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-excel-modeling-v1-1 · How It Works · Data refreshed daily, snapshot 2026-10-21.