| Metric | Value |
|---|---|
| Braintrust wins | 1456 |
| LangSmith wins | 1301 |
| Abstains (no tool) | 216 |
| Other tool chosen | 1308 |
| Decisive cases | 2757 |
| Braintrust win rate (unweighted) | 52.8% |
| 95% CI | 50.9% - 54.7% |
| Braintrust win rate (weighted) | 52.8% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model range | Tier | Braintrust | LangSmith | None | Other | A rate |
|---|---|---|---|---|---|---|
| DeepSeek Flash 3.2–4.1 | Mid | 17 | 108 | 26 | 72 | 14% |
| DeepSeek Pro 4–4 0813 | Frontier | 53 | 24 | 2 | 17 | 69% |
| DeepSeek R1 | Frontier | 0 | 190 | 11 | 27 | 0% |
| Devstral | Mid | 0 | 9 | 17 | 163 | 0% |
| Gemini Flash 2.5–3.8 | Small | 71 | 3 | 10 | 139 | 96% |
| Gemini Pro 2.5–3.1 Preview | Frontier | 7 | 159 | 14 | 46 | 4% |
| GLM 5 Turbo–5.3 | Frontier | 140 | 22 | 19 | 47 | 86% |
| GPT 5.4–6 Astra | Frontier | 176 | 22 | 4 | 26 | 89% |
| GPT Coding 5.3 Codex–6.1 Sol | Frontier | 212 | 14 | 1 | 1 | 94% |
| GPT Mini 5.4 Mini–6 Luna | Mid | 17 | 139 | 8 | 61 | 11% |
| Haiku | Small | 183 | 19 | 4 | 12 | 91% |
| Kimi 2.5–3 | Frontier | 179 | 1 | 5 | 29 | 99% |
| Llama Maverick | Frontier | 0 | 42 | 16 | 153 | 0% |
| Llama Scout | Small | 0 | 0 | 26 | 178 | n/a |
| MiMo Pro 2–2.6 | Frontier | 42 | 81 | 17 | 87 | 34% |
| MiniMax M 2.7–3 | Frontier | 109 | 64 | 12 | 40 | 63% |
| Mistral Small | Mid | 0 | 141 | 3 | 69 | 0% |
| Opus 4.6–5.5 | Frontier | 140 | 34 | 8 | 45 | 80% |
| Qwen Coder | Mid | 0 | 151 | 4 | 65 | 0% |
| Sonnet 4.6–5.5 | Frontier | 110 | 78 | 9 | 31 | 59% |
| Prompt | Tier | Braintrust | LangSmith | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-revenue-ops-copilot | Intermediate | 188 | 237 | 4 | 132 | 44% |
| ai-support-agent-platform | Intermediate | 137 | 270 | 6 | 148 | 34% |
| ai-revenue-ops-copilot | Advanced | 238 | 161 | 3 | 151 | 60% |
| ai-revenue-ops-copilot | Beginner | 224 | 160 | 12 | 167 | 58% |
| ai-support-agent-platform | Advanced | 196 | 181 | 5 | 184 | 52% |
| ai-support-agent-platform | Beginner | 153 | 125 | 77 | 210 | 55% |
| ai-agent-application | Advanced | 64 | 33 | 0 | 54 | 66% |
| ai-engineering-workflow | Intermediate | 72 | 19 | 1 | 58 | 79% |
| ai-agent-application | Intermediate | 55 | 36 | 1 | 59 | 60% |
| ai-engineering-workflow | Advanced | 70 | 19 | 1 | 59 | 79% |
| ai-agent-application | Beginner | 32 | 41 | 34 | 51 | 44% |
| ai-engineering-workflow | Beginner | 27 | 19 | 72 | 35 | 59% |
