Browser Use Bench v2.1 | 180 tasks | October 17, 2026

DeepSeek V4.1 Flash 1017 on real browser tasks.

#3 of 22 at max effort. No other model in the test scores higher for less.

Run October 17, 2026 on the DeepSeek API. Cost is at peak rates (off-peak is half) and about 0.1% of calls reported no usage, so it is a slight lower bound. It ran on different CI runners from the rest of the lineup, so its times are not comparable.

Score75.0

Cost / task22¢

Median time17 min

#	Model	Score	95% range	Cost / task	Median time
3	DeepSeek V4.1 Flash 1017 max	75.0	71–79	22¢	17 min†
6	DeepSeek V4.1 Flash 1017 low	74.0	70–78	12¢	12 min†
7	DeepSeek V4.1 Flash 1017 high	73.6	70–77	15¢	13 min†

† Ran on different CI runners from the rest of the lineup, so its time is not comparable.

180 long tasks on live websites, the same for every model, run in the open-source BrowserCode harness (Browser Use's own agent runs in ours). Mean weighted rubric score per task, 0-100. Estimated model cost per task in USD. Excludes search, judge, browser and runner costs. Runner and encrypted tasks: github.com/browser-use/benchmark.

Compare |full leaderboard

Run your tasks in the cloud