| Metric | Value |
|---|---|
| Sanity wins | 546 |
| Contentful wins | 383 |
| Abstains (no tool) | 157 |
| Other tool chosen | 1384 |
| Decisive cases | 929 |
| Sanity win rate (unweighted) | 58.8% |
| 95% CI | 55.6% - 61.9% |
| Sanity win rate (weighted) | 58.8% |
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 | 70 | 27 | 4 | 25 | 72% |
| Kimi K2.5 | Frontier | 91 | 1 | 9 | 29 | 99% |
| GPT 5.3 Codex | Frontier | 60 | 31 | 4 | 34 | 66% |
| Qwen3 Coder Next | Mid | 46 | 27 | 47 | 12 | 63% |
| Mistral Small 4 | Mid | 36 | 35 | 0 | 55 | 51% |
| DeepSeek V3.2 | Mid | 44 | 20 | 6 | 62 | 69% |
| Claude Opus 4.6 | Frontier | 44 | 16 | 0 | 72 | 73% |
| GPT 5.4 | Frontier | 43 | 17 | 0 | 72 | 72% |
| Gemini 2.5 Flash | Small | 0 | 59 | 0 | 73 | 0% |
| GLM 5 Turbo | Frontier | 36 | 9 | 43 | 43 | 80% |
| Gemini 2.5 Pro | Frontier | 18 | 21 | 2 | 91 | 46% |
| MiniMax M2.7 | Frontier | 16 | 23 | 5 | 86 | 41% |
| DeepSeek R1 0528 | Frontier | 14 | 22 | 9 | 86 | 39% |
| MiMo V2 Pro | Frontier | 9 | 23 | 4 | 96 | 28% |
| Claude Haiku 4.5 | Small | 0 | 27 | 0 | 105 | 0% |
| Devstral 2 2512 | Mid | 0 | 22 | 0 | 110 | 0% |
| Claude Sonnet 4.6 | Frontier | 19 | 0 | 0 | 113 | 100% |
| Llama 4 Scout | Small | 0 | 2 | 4 | 121 | 0% |
| Llama 4 Maverick | Frontier | 0 | 1 | 20 | 99 | 0% |
| Prompt | Tier | Sanity | Contentful | None | Other | A rate |
|---|---|---|---|---|---|---|
| documentation-site | Advanced | 119 | 179 | 14 | 91 | 40% |
| blog-platform-cms | Intermediate | 170 | 15 | 0 | 233 | 92% |
| documentation-site | Intermediate | 72 | 93 | 24 | 219 | 44% |
| blog-platform-cms | Advanced | 84 | 53 | 29 | 246 | 61% |
| blog-platform-cms | Beginner | 92 | 31 | 0 | 295 | 75% |
| documentation-site | Beginner | 9 | 12 | 90 | 300 | 43% |
