Browser Use Bench v2.1 | 180 tasks | October 29, 2026

GPT-5.6 Luna on real browser tasks.

#21 of 40 at xhigh effort. DeepSeek V4.1 Flash 1017 max scores 5.4 points higher for 1.3× less.

Score69.6
Cost / task28¢
Median time15 min
#ModelScoreCost / task
21GPT-5.6 Luna xhigh69.628¢
40GPT-5.6 Luna none31.04.0¢

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.