02 / IT · Analysis
browser-use makes 15-times token pricing the default
Developers must explicitly pass llm='bu-fast' to retain the previous model and pricing.

browser-use announced BU 4.0 as its default model, with token prices 15 times those of bu-fast for calls that omit a model setting.
According to the release notes, the change applies when a call leaves the model unspecified. For developers using browser-use in game development, that call setting determines which model and token prices apply.
The release notes set out the new default pricing alongside the setting needed to retain the previous model. Keeping the earlier pricing therefore depends on explicitly selecting bu-fast.
The setting that preserves previous pricing
The release notes specify that passing llm='bu-fast' retains the previous model and pricing. Developers must name that model in the call to keep those rates.
According to the release notes, Agent(task) and ChatBrowserUse() now default to bu-4.0. Both run through the Browser Use gateway.
The release notes recommend bu-4.0-mini as the lowest-cost option for simple tasks. They direct developers to select bu-4.0-pro for the hardest tasks.
The cost of the default change
The release notes give separate input and output token prices for BU 4.0 and bu-fast. The 15-fold price difference applies to both input and output tokens.
The release notes also state that default reasoning effort changed from medium to high. The default change therefore covers both token pricing and reasoning effort, the setting that controls how much reasoning the model uses.
The test-task cost example in the release notes reflects the changed default settings. For game developers retaining the previous model and pricing, the documented setting is llm='bu-fast'; browser-use did not disclose the test task's contents or the number of runs.