LangChain benchmarks 7 browser toolsets: Stagehand wins with 90% pass rate at $1.98 vs $83.62
LangChain · x · 2026-11-04
LangChain tested 7 toolsets on a saturated browser-use benchmark (WebVoyager, 100 tasks, Claude Sonnet 5.5) to measure token/cost efficiency.
- Stagehand tools (code mode) won: 90% pass rate, $1.98 total cost, 4.0M tokens, 17.4s median time
- browser-use: 85% but $83.62 and 41.4M tokens
- Playwright MCP: 78%, $26.71
- Browserbase MCP: 89% pass rate but 65M tokens (native) vs 81%/$4.52 (code mode)
- browse CLI: 85%, $11.97
Takeaway: tool design matters enormously — the same model can vary 40x+ in cost depending on the toolset.
More from coding & agent
- DeepSeek article outlines super-agent architecture: continuous, productive, proactive, persistent — LangChain · 2026-11-04
- Salesforce's 700-person AI org switched from Cursor to Claude Code, no mandate needed — aakashgupta · 2026-11-04
- Wrap, an AI-first Mac screen recorder, uses your action log to let an agent edit footage — therealdanvega · 2026-11-04
- Indie-Built Sparkboard Lets AI Agents Research Together and Keep Digging for Days — HankYeomans · 2026-11-04
- Binance dev-ops infected after downloading Homebrew from malicious Google search ad — HankYeomans · 2026-11-04
- Filmmaker uses Claude Opus to direct AI movie: still needs tons of reshoots for 150 minutes — MosskeepForest · 2026-11-04