BrowseComp-Plus leaderboard
- GPT-5 leads
- 80.0%
- models measured
- 11
- lead over #2
- +13.0 pts
- median
- 48.2%
Leader on BrowseComp-Plus · read October 21
On October 21, 2026, 12 of our catalog's models had a BrowseComp-Plus result, and OpenAI GPT-5 led with 80.0%, ahead of OpenAI gpt-oss-120b at 67.0%.
BrowseComp-Plus gives a model hard research questions from BrowseComp and a search tool over a fixed collection of web documents, so results do not depend on a live search engine. The number is the share of questions answered correctly. Each model is run with several retrievers; we show its best result with the standard search setup and name the retriever. Runs that bring their own agent scaffold are left out.
Role in the index
BrowseComp-Plus is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | GPT-5OpenAI | 80.0% | 38.8 | $10.00 | |
| 2 | gpt-oss-120bOpenAI | 67.0% | Not benchmarked | $0.50 | |
| 3 | o3OpenAI | 65.9% | Not benchmarked | $8.00 | |
| 4 | GLM 4.7Z.ai | 64.7% | 55.1 | $2.20 | |
| 5 | gpt-oss-20bOpenAI | 50.6% | Not benchmarked | $0.15 | |
| 6 | GLM 4.6Z.ai | 48.2% | 8.8 | $2.20 | |
| 7 | Gemini 2.5 FlashGoogle | 39.9% | Not benchmarked | $2.50 | |
| 8 | Claude Opus 4RetiredAnthropic | 36.8% | Not benchmarked | $75.00 | |
| 9 | GPT-4.1OpenAI | 36.4% | Not benchmarked | $8.00 | |
| 10 | Gemini 2.5 ProGoogle | 29.5% | 19.1 | $10.00 | |
| 11 | R1 0528DeepSeek | 16.4% | Not benchmarked | $2.18 | |
| 12 | Qwen3 32BQwen | 10.7% | Not benchmarked | $0.40 |
Only models in our catalog are listed, and each appears once with its best published run. 13 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out.