Understand this story
On November 10, 2026, browser-use 1.0 replaced the 24 browser operations in version 0.x with a persistent Python REPL and opened up direct access to the Chrome DevTools Protocol. BU Bench v2.1 results published by Browser Use show that scores for the same three models increased by 14 to 33 points, while costs fell to between 1/2.7 and 1/5.2 of their previous levels. However, because the version's default settings also changed, these improvements cannot be attributed entirely to the tool architecture overhaul. The new architecture allows agents to handle scenarios not covered by predefined operations, such as frozen pages and calls to private APIs, though specialized tools remained more effective in CAPTCHA tests. There are currently no other subsequent developments supported by factual evidence.
AI-generated from public reports
Latest development · Report
browser-use 1.0 Replaces 24 Agent Tools with a Single Python ToolDevelopments
1 confirmed developments
browser-use 1.0 replaces 24 browser actions with a single Python tool
Source:Browser Use
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 20
Browser UseFirst-partySelected
browser-use 1.0 Replaces 24 Agent Tools with a Single Python Tool
browser-use 1.0 replaces the 24 browser actions in version 0.x with a persistent Python REPL and provides raw access to the Chrome DevTools Protocol. On BU Bench v2.1, the same three models gained 14 to 33 points while costs fell to between 1/2.7 and 1/5.2 of their previous levels. However, the version's default settings also changed, so the improvements cannot be attributed entirely to the tool redesign. The new architecture allows agents to independently handle scenarios not covered by predefined actions, such as frozen pages and calls to private APIs, while specialized tools remain more effective in CAPTCHA tests.
AI Trends