Browser use: Muse Ember matches GPT-6 Luna at 3 cents per task
WenhuChen · x · 2026-11-10
Browser use ran Muse Ember on the hardest long-horizon browser benchmark, scoring 63.9/100 — on par with GPT-6 Luna medium — at roughly 3 cents per task, 2.4x cheaper, with the best score under 5 cents. Tasks spanning hundreds of browser steps now cost cents, not dollars. Wenhu Chen amplified the result, and the team congratulated Alexandr Wang: cheap models that can actually drive a browser are a big deal.
More from coding & agent
- HuggingFace SafeAgents: a safety layer between your harness and your model — blaizedsouza · 2026-11-10
- rakyll teases Agent Substrate as 'agentic compute fabric' ahead of KubeCon keynote — rakyll · 2026-11-10
- Developer uses a 1.9B decision model on a laptop to flag dirty restaurant tables in real time — JosephJacks_ · 2026-11-10
- Microsoft FORGE Lab uses agentic AI tools to find memory-safety bugs in Rust core libs — jedisct1 · 2026-11-10
- CTM-AI: open-source multi-agent framework inspired by consciousness theory, NeurIPS 2026 paper — pliang279 · 2026-11-10
- Andrew Ng's AI Dev Conference Hits NYC Next November, Tickets From $599 — DeepLearningAI · 2026-11-10