| Metric | Value |
|---|---|
| Langfuse wins | 1080 |
| LangSmith wins | 2103 |
| Abstains (no tool) | 122 |
| Other tool chosen | 584 |
| Decisive cases | 3183 |
| Langfuse win rate (unweighted) | 33.9% |
| 95% CI | 32.3% - 35.6% |
| Langfuse win rate (weighted) | 33.9% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | Langfuse | LangSmith | None | Other | A rate |
|---|---|---|---|---|---|---|
| GPT 5.3 Codex | Frontier | 181 | 23 | 0 | 0 | 89% |
| Claude Sonnet 4.6 | Frontier | 53 | 145 | 6 | 0 | 27% |
| GPT 5.4 Mini | Mid | 43 | 152 | 5 | 4 | 22% |
| Gemini 2.5 Pro | Frontier | 6 | 186 | 7 | 5 | 3% |
| DeepSeek R1 0528 | Frontier | 0 | 183 | 2 | 19 | 0% |
| Claude Haiku 4.5 | Small | 129 | 52 | 2 | 17 | 71% |
| Qwen3 Coder Next | Mid | 75 | 100 | 2 | 26 | 43% |
| Mistral Small 4 | Mid | 10 | 157 | 3 | 29 | 6% |
| GPT 5.4 | Frontier | 119 | 13 | 0 | 0 | 90% |
| Claude Opus 4.6 | Frontier | 110 | 22 | 0 | 0 | 83% |
| GLM 5 Turbo | Frontier | 12 | 120 | 0 | 0 | 9% |
| Llama 4 Maverick | Frontier | 17 | 112 | 12 | 62 | 13% |
| DeepSeek V3.2 | Mid | 7 | 120 | 0 | 5 | 6% |
| MiMo V2 Pro | Frontier | 1 | 125 | 2 | 4 | 1% |
| Kimi K2.5 | Frontier | 50 | 65 | 4 | 0 | 43% |
| MiniMax M2.7 | Frontier | 1 | 105 | 3 | 21 | 1% |
| GLM 5.2 | Frontier | 30 | 42 | 0 | 0 | 42% |
| MiniMax M3 | Frontier | 42 | 28 | 1 | 1 | 60% |
| Kimi K2.7 Code | Frontier | 35 | 35 | 0 | 1 | 50% |
| GPT 5.5 | Frontier | 39 | 30 | 2 | 1 | 57% |
| DeepSeek V4 Pro | Frontier | 10 | 59 | 1 | 2 | 14% |
| Gemini 3.5 Flash | Small | 8 | 60 | 1 | 3 | 12% |
| DeepSeek V4 Flash | Mid | 24 | 43 | 4 | 0 | 36% |
| Claude Opus 4.8 | Frontier | 32 | 34 | 6 | 0 | 48% |
| MiMo V2.5 Pro | Frontier | 9 | 56 | 5 | 2 | 14% |
| Llama 4 Scout | Small | 36 | 0 | 27 | 129 | 100% |
| Devstral 2 2512 | Mid | 0 | 32 | 26 | 127 | 0% |
| Gemini 2.5 Flash | Small | 1 | 4 | 1 | 126 | 20% |
| Prompt | Tier | Langfuse | LangSmith | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-support-agent-platform | Intermediate | 103 | 352 | 1 | 79 | 23% |
| ai-revenue-ops-copilot | Intermediate | 115 | 330 | 1 | 75 | 26% |
| ai-support-agent-platform | Beginner | 202 | 235 | 16 | 78 | 46% |
| ai-revenue-ops-copilot | Advanced | 124 | 308 | 3 | 91 | 29% |
| ai-support-agent-platform | Advanced | 155 | 274 | 1 | 103 | 36% |
| ai-revenue-ops-copilot | Beginner | 135 | 285 | 32 | 84 | 32% |
| ai-agent-application | Advanced | 40 | 69 | 0 | 8 | 37% |
| ai-agent-application | Intermediate | 20 | 86 | 1 | 11 | 19% |
| ai-agent-application | Beginner | 23 | 75 | 19 | 2 | 23% |
| ai-engineering-workflow | Intermediate | 61 | 36 | 0 | 21 | 63% |
| ai-engineering-workflow | Advanced | 57 | 35 | 0 | 24 | 62% |
| ai-engineering-workflow | Beginner | 45 | 18 | 48 | 8 | 71% |
