Wordsmith Model Bench v1 - Polish: leaderboard
Wordsmith October 15, 2026 v1 pilot on unpublished manuscripts. Cross-maker AI judges never grade their own maker. Exact source effort settings are retained. Scores reward correct useful editorial work; manuscripts, prompts and checklist are withheld. Metric: Editorial score (0–100).
Metric: Editorial score (0–100). Source: wordsmith.page. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon (High) | 99.48 | #8 (Gemini 4 Argon) |
| 2 | Gemini 3.1 Pro (Preview) (High) | 99.25 | #61 (Gemini 3.1 Pro (Preview)) |
| 3 | Gemini 3.8 Flash (High) | 98.93 | #23 (Gemini 3.8 Flash) |
| 4 | Claude Opus 5.5 (Low) | 98.67 | #5 (Claude Opus 5.5) |
| 5 | Claude Sonnet 5.5 (High) | 98.09 | #12 (Claude Sonnet 5.5) |
| 6 | Claude Fable 5.1 (High) | 98.02 | #9 (Claude Fable 5.1) |
| 7 | Claude Opus 5.5 (High) | 97.88 | #5 (Claude Opus 5.5) |
| 8 | GPT-6.1 Sol (Medium) | 97.75 | #7 (GPT-6.1 Sol) |
| 9 | GPT-6.1 Sol (High) | 97.51 | #7 (GPT-6.1 Sol) |
| 10 | GPT-6 Astra (High) | 96.23 | #6 (GPT-6 Astra) |
| 11 | Claude Haiku 5.5 (High) | 94.89 | #40 (Claude Haiku 5.5) |
| 12 | GPT-5.6 Terra (High) | 91.28 | #38 (GPT-5.6 Terra) |
No result here: #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=wordsmith-model-bench-v1-polish · How It Works · Data refreshed daily, snapshot 2026-10-26.