Wordsmith Model Bench v1 - Craft: 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Wordsmith October 15, 2026 v1 pilot on unpublished manuscripts.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: Around 2028. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (High) | 69.09 | #9 (Claude Fable 5.1) |
| 2 | Claude Opus 5.5 (High) | 67.32 | #5 (Claude Opus 5.5) |
| 3 | Gemini 4 Argon (High) | 62.04 | #8 (Gemini 4 Argon) |
| 4 | GPT-6.1 Sol (High) | 57.59 | #7 (GPT-6.1 Sol) |
| 5 | Claude Sonnet 5.5 (High) | 57.38 | #12 (Claude Sonnet 5.5) |
| 6 | Gemini 3.8 Flash (High) | 57.24 | #23 (Gemini 3.8 Flash) |
| 7 | Claude Haiku 5.5 (High) | 55.84 | #40 (Claude Haiku 5.5) |
| 8 | GPT-6 Astra (High) | 54.65 | #6 (GPT-6 Astra) |
| 9 | GPT-5.6 Terra (High) | 51.06 | #38 (GPT-5.6 Terra) |
| 10 | Gemini 3.1 Pro (Preview) (High) | 48.53 | #61 (Gemini 3.1 Pro (Preview)) |
| 11 | GPT-6.1 Sol (Medium) | 48.47 | #7 (GPT-6.1 Sol) |
| 12 | Claude Opus 5.5 (Low) | 34.4 | #5 (Claude Opus 5.5) |
No result here: #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=wordsmith-model-bench-v1-craft · How It Works · Data refreshed daily, snapshot 2026-10-26.