Custom Models
Web Agent
LLMs
Models trained specifically for browser automation. Higher accuracy, lower latency, fraction of the cost.
State-of-the-art accuracy
Purpose-built for browser tasks. Outperforms general-purpose frontier models at a fraction of the latency and cost.
Read Blog Post →4x faster execution
68 seconds per task vs 225-330s for Gemini, Claude, and OpenAI computer use models. Optimized output parsing and batched caching.
Fraction of the cost
53 tasks per dollar vs 2 for Sonnet 4.5. Purpose-built models eliminate the overhead of general-purpose frontier LLMs.
Trained on the browser. Priced like it.
Browser Use Cloud (bu-ultra): 78% of tasks solved. 14.7 points ahead of OSS + ChatBrowserUse-2, the next best.
78%
63.3%
62%
59.3%
59%
52.4%
37%
35.2%
Browser Use Cloud (bu-ultra)
OSS + ChatBrowserUse-2
claude-opus-4-6
gemini-3-1-pro
claude-sonnet-4-6
gpt-5
gpt-5-mini
gemini-2.5-flash
ProviderAccuracy
Browser Use Cloud (bu-ultra)78%
OSS + ChatBrowserUse-263.3%
claude-opus-4-662%
gemini-3-1-pro59.3%
claude-sonnet-4-659%
gpt-552.4%
gpt-5-mini37%
gemini-2.5-flash35.2%