Skip to content
Lusor
PricingStart for freeLog in

Benchmarks

Lusor benchmarks: phone agents on public tasks, every run recorded

Each benchmark uses fixed public tasks, grades every Lusor run and says where every other number comes from.

Success rate (pass@1) · higher is better · scale 0–100%

  • LusorMobilerun’s grader100%
  • Mobilerunreported by its team100%
  • agent-qareported by its team100%
  • FluizAIreported by its team100%
  • Artemisreported by its team99.1%

+2 more in the full table

AndroidWorld

This run didn’t need a flagship closed model as its actor. Lusor’s production agent, with open-weight DeepSeek V4.1 Flash choosing every action, passed all 116 tasks on Mobilerun’s grader and 114 on Google’s unmodified one, a median 5.2× faster than Mobilerun’s published per-task times. The tasks are Google Research’s 116, in real Android apps, graded by the benchmark’s own code.

Mobilerun’s grader
116/116
faster, median per task
5.2×
for all 116 tasks
$2.95
7 systems116 tasks1 trial per task
Read the benchmark →

Every AndroidWorld figure here, ours included, is reported by the team that ran it; the AndroidWorld community sheet (exported Nov 10, 2026) does not verify entries. We publish all 116 of our runs so ours can be checked.