BrowseComp — OpenAI browsing-agent benchmark, tools-enabled mode; distinct from other tool/agent variants
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Kimi K2 (Reasoning) | Moonshot | 60.2 | 2 | 2026-06-12 |
| 2 | GPT-5 | OpenAI | 54.9 | 1 | 2026-06-12 |
| 3 | DeepSeek-V3.2 | DeepSeek | 40.1 | 1 | 2026-06-12 |
| 4 | Claude Sonnet 4.5 | Anthropic | 24.1 | 1 | 2026-06-12 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.