| Metric | Value |
|---|---|
| Portkey wins | 34 |
| LiteLLM wins | 45 |
| Abstains (no tool) | 1 |
| Other tool chosen | 30 |
| Decisive cases | 79 |
| Portkey win rate (unweighted) | 43.0% |
| 95% CI | 32.7% - 54.0% |
| Portkey win rate (weighted) | 43.0% |
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 | 6 | 0 | 0 | 0 | 100% |
| DeepSeek R1 0528 | Frontier | 5 | 1 | 0 | 0 | 83% |
| DeepSeek V4 Flash | Mid | 2 | 4 | 0 | 0 | 33% |
| Gemini 3.5 Flash | Small | 1 | 5 | 0 | 0 | 17% |
| MiniMax M3 | Frontier | 5 | 0 | 0 | 1 | 100% |
| Devstral 2 2512 | Mid | 3 | 2 | 0 | 1 | 60% |
| Claude Sonnet 4.6 | Frontier | 0 | 5 | 0 | 1 | 0% |
| GPT 5.4 Mini | Mid | 0 | 5 | 0 | 1 | 0% |
| Kimi K2.7 Code | Frontier | 0 | 5 | 0 | 0 | 0% |
| Llama 4 Maverick | Frontier | 0 | 5 | 1 | 0 | 0% |
| DeepSeek V4 Pro | Frontier | 4 | 0 | 0 | 2 | 100% |
| GPT 5.3 Codex | Frontier | 4 | 0 | 0 | 2 | 100% |
| MiMo V2.5 Pro | Frontier | 0 | 4 | 0 | 2 | 0% |
| GLM 5.2 | Frontier | 3 | 0 | 0 | 3 | 100% |
| Claude Opus 4.8 | Frontier | 1 | 2 | 0 | 3 | 33% |
| GPT 5.5 | Frontier | 0 | 3 | 0 | 2 | 0% |
| Mistral Small 4 | Mid | 0 | 3 | 0 | 2 | 0% |
| Claude Haiku 4.5 | Small | 0 | 1 | 0 | 4 | 0% |
| Llama 4 Scout | Small | 0 | 0 | 0 | 1 | n/a |
| Qwen3 Coder Next | Mid | 0 | 0 | 0 | 5 | n/a |
| Prompt | Tier | Portkey | LiteLLM | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-engineering-workflow | Intermediate | 7 | 10 | 0 | 3 | 41% |
| ai-engineering-workflow | Advanced | 9 | 6 | 0 | 1 | 60% |
| ai-agent-application | Advanced | 8 | 7 | 0 | 3 | 53% |
| ai-agent-application | Intermediate | 5 | 9 | 0 | 4 | 36% |
| ai-agent-application | Beginner | 2 | 9 | 0 | 8 | 18% |
| ai-engineering-workflow | Beginner | 3 | 4 | 1 | 11 | 43% |
