| Metric | Value |
|---|---|
| LangSmith wins | 1341 |
| Langfuse wins | 646 |
| Abstains (no tool) | 45 |
| Other tool chosen | 441 |
| Decisive cases | 1987 |
| LangSmith win rate (unweighted) | 67.5% |
| 95% CI | 65.4% - 69.5% |
| LangSmith win rate (weighted) | 67.5% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | LangSmith | Langfuse | None | Other | A rate |
|---|---|---|---|---|---|---|
| GLM 5 Turbo | Frontier | 120 | 12 | 0 | 0 | 91% |
| Claude Sonnet 4.6 | Frontier | 102 | 30 | 0 | 0 | 77% |
| Claude Opus 4.6 | Frontier | 22 | 110 | 0 | 0 | 17% |
| GPT 5.3 Codex | Frontier | 14 | 118 | 0 | 0 | 11% |
| GPT 5.4 | Frontier | 13 | 119 | 0 | 0 | 10% |
| GPT 5.4 Mini | Mid | 105 | 23 | 1 | 3 | 82% |
| DeepSeek V3.2 | Mid | 120 | 7 | 0 | 5 | 94% |
| MiMo V2 Pro | Frontier | 125 | 1 | 2 | 4 | 99% |
| Gemini 2.5 Pro | Frontier | 122 | 1 | 6 | 3 | 99% |
| DeepSeek R1 0528 | Frontier | 118 | 0 | 1 | 13 | 100% |
| Mistral Small 4 | Mid | 112 | 4 | 0 | 13 | 97% |
| Qwen3 Coder Next | Mid | 61 | 55 | 1 | 14 | 53% |
| Claude Haiku 4.5 | Small | 37 | 79 | 0 | 13 | 32% |
| Kimi K2.5 | Frontier | 65 | 50 | 4 | 0 | 57% |
| MiniMax M2.7 | Frontier | 105 | 1 | 3 | 21 | 99% |
| Llama 4 Maverick | Frontier | 75 | 11 | 0 | 46 | 87% |
| Llama 4 Scout | Small | 0 | 24 | 11 | 90 | 0% |
| Devstral 2 2512 | Mid | 21 | 0 | 15 | 90 | 100% |
| Gemini 2.5 Flash | Small | 4 | 1 | 1 | 126 | 80% |
| Prompt | Tier | LangSmith | Langfuse | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-support-agent-platform | Intermediate | 264 | 81 | 1 | 70 | 77% |
| ai-revenue-ops-copilot | Intermediate | 251 | 87 | 1 | 64 | 74% |
| ai-support-agent-platform | Beginner | 174 | 159 | 11 | 69 | 52% |
| ai-revenue-ops-copilot | Advanced | 232 | 96 | 2 | 79 | 71% |
| ai-support-agent-platform | Advanced | 206 | 118 | 1 | 90 | 64% |
| ai-revenue-ops-copilot | Beginner | 214 | 105 | 29 | 69 | 67% |
