| Metric | Value |
|---|---|
| Portkey wins | 204 |
| LiteLLM wins | 277 |
| Abstains (no tool) | 6 |
| Other tool chosen | 164 |
| Decisive cases | 481 |
| Portkey win rate (unweighted) | 42.4% |
| 95% CI | 38.1% - 46.9% |
| Portkey win rate (weighted) | 42.4% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | Portkey | LiteLLM | None | Other | A rate |
|---|---|---|---|---|---|---|
| Gemini 2.5 Pro | Frontier | 35 | 1 | 0 | 0 | 97% |
| DeepSeek V4 Pro | Frontier | 21 | 13 | 0 | 2 | 62% |
| DeepSeek V4 Flash | Mid | 16 | 18 | 0 | 2 | 47% |
| DeepSeek R1 0528 | Frontier | 26 | 5 | 1 | 4 | 84% |
| Gemini 3.5 Flash | Small | 10 | 21 | 0 | 5 | 32% |
| Llama 4 Maverick | Frontier | 5 | 26 | 1 | 4 | 16% |
| Claude Sonnet 4.6 | Frontier | 2 | 28 | 0 | 6 | 7% |
| MiniMax M3 | Frontier | 25 | 4 | 0 | 6 | 86% |
| MiMo V2.5 Pro | Frontier | 5 | 23 | 0 | 8 | 18% |
| Devstral 2 2512 | Mid | 15 | 12 | 0 | 1 | 56% |
| GPT 5.5 | Frontier | 4 | 21 | 0 | 8 | 16% |
| GPT 5.3 Codex | Frontier | 19 | 5 | 0 | 11 | 79% |
| GLM 5.2 | Frontier | 17 | 6 | 0 | 13 | 74% |
| Kimi K2.7 Code | Frontier | 1 | 21 | 0 | 3 | 5% |
| GPT 5.4 Mini | Mid | 0 | 22 | 1 | 13 | 0% |
| Claude Opus 4.8 | Frontier | 2 | 18 | 0 | 16 | 10% |
| Mistral Small 4 | Mid | 0 | 19 | 3 | 4 | 0% |
| Claude Haiku 4.5 | Small | 0 | 13 | 0 | 15 | 0% |
| Llama 4 Scout | Small | 1 | 0 | 0 | 13 | 100% |
| Qwen3 Coder Next | Mid | 0 | 1 | 0 | 30 | 0% |
| Prompt | Tier | Portkey | LiteLLM | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-agent-application | Advanced | 43 | 50 | 0 | 13 | 46% |
| ai-engineering-workflow | Intermediate | 44 | 48 | 0 | 12 | 48% |
| ai-engineering-workflow | Advanced | 56 | 29 | 0 | 19 | 66% |
| ai-agent-application | Intermediate | 25 | 60 | 1 | 25 | 29% |
| ai-agent-application | Beginner | 12 | 71 | 1 | 30 | 14% |
| ai-engineering-workflow | Beginner | 24 | 19 | 4 | 65 | 56% |
