| Metric | Value |
|---|---|
| Sanity wins | 195 |
| Contentful wins | 145 |
| Abstains (no tool) | 56 |
| Other tool chosen | 500 |
| Decisive cases | 340 |
| Sanity win rate (unweighted) | 57.4% |
| 95% CI | 52.0% - 62.5% |
| Sanity win rate (weighted) | 57.4% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | Sanity | Contentful | None | Other | A rate |
|---|---|---|---|---|---|---|
| GPT 5.4 Mini | Mid | 25 | 10 | 1 | 11 | 71% |
| Kimi K2.5 | Frontier | 32 | 0 | 2 | 13 | 100% |
| GPT 5.3 Codex | Frontier | 22 | 10 | 2 | 13 | 69% |
| Qwen3 Coder Next | Mid | 16 | 12 | 15 | 5 | 57% |
| Claude Opus 4.6 | Frontier | 16 | 8 | 0 | 24 | 67% |
| DeepSeek V3.2 | Mid | 14 | 9 | 2 | 23 | 61% |
| Mistral Small 4 | Mid | 12 | 11 | 0 | 21 | 52% |
| GPT 5.4 | Frontier | 15 | 7 | 0 | 26 | 68% |
| Gemini 2.5 Flash | Small | 0 | 21 | 0 | 27 | 0% |
| GLM 5 Turbo | Frontier | 14 | 3 | 17 | 14 | 82% |
| DeepSeek R1 0528 | Frontier | 5 | 12 | 3 | 28 | 29% |
| MiniMax M2.7 | Frontier | 8 | 5 | 4 | 31 | 62% |
| Gemini 2.5 Pro | Frontier | 5 | 8 | 1 | 34 | 38% |
| MiMo V2 Pro | Frontier | 3 | 8 | 0 | 37 | 27% |
| Claude Haiku 4.5 | Small | 0 | 11 | 0 | 37 | 0% |
| Claude Sonnet 4.6 | Frontier | 8 | 0 | 0 | 40 | 100% |
| Devstral 2 2512 | Mid | 0 | 8 | 0 | 40 | 0% |
| Llama 4 Maverick | Frontier | 0 | 1 | 8 | 33 | 0% |
| Llama 4 Scout | Small | 0 | 1 | 1 | 43 | 0% |
| Prompt | Tier | Sanity | Contentful | None | Other | A rate |
|---|---|---|---|---|---|---|
| documentation-site | Advanced | 47 | 65 | 4 | 31 | 42% |
| blog-platform-cms | Intermediate | 63 | 7 | 0 | 82 | 90% |
| documentation-site | Intermediate | 25 | 38 | 12 | 71 | 40% |
| blog-platform-cms | Beginner | 34 | 13 | 0 | 105 | 72% |
| blog-platform-cms | Advanced | 24 | 19 | 8 | 100 | 56% |
| documentation-site | Beginner | 2 | 3 | 32 | 111 | 40% |
