Brave Search (count=10)
Stub model card so this search API can appear on the OB News Websearch Official Benchmark leaderboard. There are no weights here.
- Band: Web search
- Endpoint:
POST /res/v1/llm/context - Config:
count=10 - Docs: https://api-dashboard.search.brave.com/documentation/services/llm-context
- Live board: https://brave.com/search/api/
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Evaluation results
- openbenchmarks/OB-News-Websearch
- Accuracy View evaluation resultssource
POST /res/v1/llm/context · count=10; $ / 1k correct=$5.32; private held-out 300; extract gpt-5.6-terra medium; judge claude-opus-5 Bedrock94 * - Web Search View evaluation resultssource POST /res/v1/llm/context · count=10; $ / 1k correct=$5.32; private held-out 300; extract gpt-5.6-terra medium; judge claude-opus-5 Bedrock94 *
- Ar At 1 View evaluation resultssource POST /res/v1/llm/context · count=10; $ / 1k correct=$5.32; private held-out 300; extract gpt-5.6-terra medium; judge claude-opus-5 Bedrock81 *