| Metric | Value |
|---|---|
| Tavily wins | 267 |
| SerpAPI wins | 47 |
| Abstains (no tool) | 1 |
| Other tool chosen | 31 |
| Decisive cases | 314 |
| Tavily win rate (unweighted) | 85.0% |
| 95% CI | 80.7% - 88.6% |
| Tavily win rate (weighted) | 85.0% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | Tavily | SerpAPI | None | Other | A rate |
|---|---|---|---|---|---|---|
| Claude Opus 4.8 | Frontier | 21 | 0 | 0 | 0 | 100% |
| Claude Sonnet 4.6 | Frontier | 21 | 0 | 0 | 0 | 100% |
| DeepSeek V4 Flash | Mid | 21 | 0 | 0 | 0 | 100% |
| GLM 5.2 | Frontier | 21 | 0 | 0 | 0 | 100% |
| GPT 5.3 Codex | Frontier | 21 | 0 | 0 | 0 | 100% |
| GPT 5.5 | Frontier | 21 | 0 | 0 | 0 | 100% |
| Kimi K2.7 Code | Frontier | 21 | 0 | 0 | 0 | 100% |
| Mistral Small 4 | Mid | 12 | 9 | 0 | 0 | 57% |
| GPT 5.4 Mini | Mid | 20 | 0 | 0 | 1 | 100% |
| MiniMax M3 | Frontier | 20 | 0 | 0 | 0 | 100% |
| DeepSeek V4 Pro | Frontier | 19 | 1 | 0 | 1 | 95% |
| DeepSeek R1 0528 | Frontier | 0 | 17 | 1 | 3 | 0% |
| Gemini 3.5 Flash | Small | 16 | 0 | 0 | 0 | 100% |
| MiMo V2.5 Pro | Frontier | 15 | 1 | 0 | 4 | 94% |
| Claude Haiku 4.5 | Small | 9 | 5 | 0 | 0 | 64% |
| Devstral 2 2512 | Mid | 0 | 13 | 0 | 0 | 0% |
| Qwen3 Coder Next | Mid | 8 | 0 | 0 | 0 | 100% |
| Llama 4 Maverick | Frontier | 1 | 1 | 0 | 10 | 50% |
| Llama 4 Scout | Small | 0 | 0 | 0 | 12 | n/a |
| Prompt | Tier | Tavily | SerpAPI | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-agent-application | Beginner | 95 | 12 | 0 | 10 | 89% |
| ai-agent-application | Intermediate | 82 | 22 | 1 | 10 | 79% |
| ai-agent-application | Advanced | 90 | 13 | 0 | 11 | 87% |
