| Metric | Value |
|---|---|
| LangSmith wins | 2170 |
| Langfuse wins | 1119 |
| Abstains (no tool) | 129 |
| Other tool chosen | 587 |
| Decisive cases | 3289 |
| LangSmith win rate (unweighted) | 66.0% |
| 95% CI | 64.3% - 67.6% |
| LangSmith win rate (weighted) | 66.0% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | LangSmith | Langfuse | None | Other | A rate |
|---|---|---|---|---|---|---|
| GPT 5.3 Codex | Frontier | 23 | 187 | 0 | 0 | 11% |
| Claude Sonnet 4.6 | Frontier | 149 | 54 | 7 | 0 | 73% |
| GPT 5.4 Mini | Mid | 156 | 45 | 5 | 4 | 78% |
| Gemini 2.5 Pro | Frontier | 192 | 7 | 6 | 5 | 96% |
| DeepSeek R1 0528 | Frontier | 188 | 0 | 2 | 20 | 100% |
| Claude Haiku 4.5 | Small | 54 | 132 | 3 | 17 | 29% |
| Qwen3 Coder Next | Mid | 103 | 77 | 2 | 27 | 57% |
| Mistral Small 4 | Mid | 159 | 10 | 3 | 33 | 94% |
| Llama 4 Maverick | Frontier | 113 | 19 | 14 | 63 | 86% |
| GLM 5 Turbo | Frontier | 114 | 12 | 0 | 0 | 90% |
| Claude Opus 4.6 | Frontier | 20 | 106 | 0 | 0 | 16% |
| GPT 5.4 | Frontier | 12 | 114 | 0 | 0 | 10% |
| DeepSeek V3.2 | Mid | 114 | 7 | 0 | 5 | 94% |
| MiMo V2 Pro | Frontier | 119 | 1 | 2 | 4 | 99% |
| Kimi K2.5 | Frontier | 62 | 48 | 4 | 0 | 56% |
| MiniMax M2.7 | Frontier | 100 | 1 | 3 | 20 | 99% |
| GLM 5.2 | Frontier | 51 | 33 | 0 | 0 | 61% |
| Kimi K2.7 Code | Frontier | 42 | 40 | 0 | 1 | 51% |
| MiniMax M3 | Frontier | 35 | 47 | 1 | 1 | 43% |
| DeepSeek V4 Pro | Frontier | 65 | 16 | 1 | 2 | 80% |
| GPT 5.5 | Frontier | 37 | 43 | 3 | 1 | 46% |
| Gemini 3.5 Flash | Small | 70 | 9 | 1 | 4 | 89% |
| DeepSeek V4 Flash | Mid | 51 | 28 | 4 | 0 | 65% |
| Claude Opus 4.8 | Frontier | 41 | 36 | 7 | 0 | 53% |
| MiMo V2.5 Pro | Frontier | 66 | 10 | 6 | 2 | 87% |
| Llama 4 Scout | Small | 0 | 36 | 29 | 133 | 0% |
| Devstral 2 2512 | Mid | 30 | 0 | 25 | 125 | 100% |
| Gemini 2.5 Flash | Small | 4 | 1 | 1 | 120 | 80% |
| Prompt | Tier | LangSmith | Langfuse | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-support-agent-platform | Intermediate | 355 | 102 | 1 | 77 | 78% |
| ai-revenue-ops-copilot | Intermediate | 330 | 115 | 1 | 76 | 74% |
| ai-support-agent-platform | Beginner | 241 | 199 | 17 | 74 | 55% |
| ai-revenue-ops-copilot | Advanced | 313 | 123 | 2 | 89 | 72% |
| ai-support-agent-platform | Advanced | 275 | 155 | 1 | 102 | 64% |
| ai-revenue-ops-copilot | Beginner | 284 | 137 | 30 | 85 | 67% |
| ai-agent-application | Advanced | 79 | 47 | 0 | 10 | 63% |
| ai-agent-application | Intermediate | 98 | 26 | 1 | 12 | 79% |
| ai-agent-application | Beginner | 89 | 25 | 21 | 3 | 78% |
| ai-engineering-workflow | Intermediate | 43 | 71 | 0 | 23 | 38% |
| ai-engineering-workflow | Advanced | 40 | 68 | 0 | 27 | 37% |
| ai-engineering-workflow | Beginner | 23 | 51 | 55 | 9 | 31% |
