BrowseComp

Can the AI browse the web and track down hard-to-find answers? Higher is better.

GPT-5.6 Sol by OpenAI holds the best BrowseComp result among tracked models, at 92.2% (released Jun 26 2026). Higher scores are better on this benchmark.

1. GPT-5.6 Sol (OpenAI) — 92.2%; 2. GPT-6 Astra (OpenAI) — 91.5%; 3. Kimi K3 (Moonshot AI) — 91.2%; 4. Claude Opus 5 (Anthropic) — 90.8%; 5. GPT-5.5-Pro (OpenAI) — 90.1%.

28 tracked models have a published BrowseComp score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.

Qwen3.5 by Qwen is the highest-ranked model with downloadable weights on BrowseComp, scoring 69% at rank 23 overall.