Leading: Portkey (63.0%)
| Metric | Value |
|---|---|
| OpenRouter wins | 20 |
| Portkey wins | 34 |
| Abstains (no tool) | 1 |
| Other tool chosen | 55 |
| Decisive cases | 54 |
| OpenRouter win rate (unweighted) | 37.0% |
| 95% CI | 25.4% - 50.4% |
| OpenRouter win rate (weighted) | 37.0% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | OpenRouter | Portkey | None | Other | A rate |
|---|---|---|---|---|---|---|
| GLM 5.2 | Frontier | 3 | 3 | 0 | 0 | 50% |
| DeepSeek V4 Pro | Frontier | 2 | 4 | 0 | 0 | 33% |
| GPT 5.3 Codex | Frontier | 2 | 4 | 0 | 0 | 33% |
| MiniMax M3 | Frontier | 1 | 5 | 0 | 0 | 17% |
| Gemini 2.5 Pro | Frontier | 0 | 6 | 0 | 0 | 0% |
| DeepSeek R1 0528 | Frontier | 0 | 5 | 0 | 1 | 0% |
| Claude Opus 4.8 | Frontier | 3 | 1 | 0 | 2 | 75% |
| Devstral 2 2512 | Mid | 0 | 3 | 0 | 3 | 0% |
| GPT 5.5 | Frontier | 2 | 0 | 0 | 3 | 100% |
| MiMo V2.5 Pro | Frontier | 2 | 0 | 0 | 4 | 100% |
| Qwen3 Coder Next | Mid | 2 | 0 | 0 | 3 | 100% |
| DeepSeek V4 Flash | Mid | 0 | 2 | 0 | 4 | 0% |
| Claude Haiku 4.5 | Small | 1 | 0 | 0 | 4 | 100% |
| Claude Sonnet 4.6 | Frontier | 1 | 0 | 0 | 5 | 100% |
| GPT 5.4 Mini | Mid | 1 | 0 | 0 | 5 | 100% |
| Gemini 3.5 Flash | Small | 0 | 1 | 0 | 5 | 0% |
| Kimi K2.7 Code | Frontier | 0 | 0 | 0 | 5 | n/a |
| Llama 4 Maverick | Frontier | 0 | 0 | 1 | 5 | n/a |
| Llama 4 Scout | Small | 0 | 0 | 0 | 1 | n/a |
| Mistral Small 4 | Mid | 0 | 0 | 0 | 5 | n/a |
| Prompt | Tier | OpenRouter | Portkey | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-engineering-workflow | Beginner | 11 | 3 | 1 | 4 | 79% |
| ai-agent-application | Intermediate | 4 | 5 | 0 | 9 | 44% |
| ai-engineering-workflow | Advanced | 0 | 9 | 0 | 7 | 0% |
| ai-agent-application | Advanced | 0 | 8 | 0 | 10 | 0% |
| ai-agent-application | Beginner | 5 | 2 | 0 | 12 | 71% |
| ai-engineering-workflow | Intermediate | 0 | 7 | 0 | 13 | 0% |
