| Metric | Value |
|---|---|
| pytest wins | 54 |
| Playwright wins | 210 |
| Abstains (no tool) | 145 |
| Other tool chosen | 69 |
| Decisive cases | 264 |
| pytest win rate (unweighted) | 20.5% |
| 95% CI | 16.0% - 25.7% |
| pytest win rate (weighted) | 20.5% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model range | Tier | pytest | Playwright | None | Other | A rate |
|---|---|---|---|---|---|---|
| DeepSeek Flash 3.2–4.1 | Mid | 3 | 12 | 2 | 8 | 20% |
| DeepSeek Pro 4–4 0813 | Frontier | 1 | 16 | 8 | 0 | 6% |
| DeepSeek R1 | Frontier | 0 | 0 | 20 | 2 | n/a |
| Devstral | Mid | 5 | 0 | 2 | 14 | 100% |
| Gemini Flash 2.5–3.8 | Small | 0 | 22 | 5 | 0 | 0% |
| Gemini Pro 2.5–3.1 Preview | Frontier | 0 | 13 | 6 | 7 | 0% |
| GLM 5 Turbo–5.3 | Frontier | 0 | 22 | 5 | 0 | 0% |
| GPT 5.4–6 Astra | Frontier | 0 | 18 | 5 | 0 | 0% |
| GPT Coding 5.3 Codex–6.1 Sol | Frontier | 2 | 16 | 6 | 0 | 11% |
| GPT Mini 5.4 Mini–6 Luna | Mid | 1 | 14 | 11 | 1 | 7% |
| Haiku | Small | 6 | 0 | 12 | 0 | 100% |
| Kimi 2.5–3 | Frontier | 0 | 11 | 11 | 1 | 0% |
| Llama Maverick | Frontier | 26 | 0 | 1 | 0 | 100% |
| Llama Scout | Small | 5 | 0 | 1 | 20 | 100% |
| MiMo Pro 2–2.6 | Frontier | 3 | 6 | 15 | 1 | 33% |
| MiniMax M 2.7–3 | Frontier | 0 | 25 | 1 | 0 | 0% |
| Mistral Small | Mid | 2 | 2 | 4 | 7 | 50% |
| Opus 4.6–5.5 | Frontier | 0 | 20 | 7 | 0 | 0% |
| Qwen Coder | Mid | 0 | 3 | 14 | 3 | 0% |
| Sonnet 4.6–5.5 | Frontier | 0 | 10 | 9 | 5 | 0% |
| Prompt | Tier | pytest | Playwright | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-engineering-workflow | Intermediate | 25 | 97 | 22 | 23 | 20% |
| ai-engineering-workflow | Advanced | 14 | 72 | 24 | 23 | 16% |
| ai-engineering-workflow | Beginner | 15 | 41 | 99 | 23 | 27% |
