Atlas

Benchmarks

Cua Driver Bench v2 (Windows, background)

Cua

Cua's Cua Driver Bench v2 on Windows 11 while a simulated user keeps typing in a window in front: eight of its ten desktop tasks, done only through a computer-use driver's tools. A trial passes when the task's checker confirms the apps' final state, no route out of the GUI was used and none of eight disturbances reached the user; each task runs five seeded trials and the rate is passes over trials.

Models
3
Top score
27.5
Leader
Claude Sonnet 5.5
Updated
—
Resolution
—
Saturation
—

No signal or difficulty estimate. Held out of the fit by design.

3 dated models · 2 frontier modelsClosed weightsOpen weights
Data: Scale AIatlas.village.dev
#ModelSettingScoreSource
1Claude Sonnet 5.5— · Cua Driver Bench on AgentEnv; Claude Code 2.1.280 with no --effort passed and its built-in tools limited to Skill and Read; trycua's cua-driver MCP tools alone; 45 turns and 360 seconds per trial
27.5
2Claude Opus 5.5— · Cua Driver Bench on AgentEnv; Claude Code 2.1.280 with no --effort passed and its built-in tools limited to Skill and Read; trycua's cua-driver MCP tools alone; 45 turns and 360 seconds per trial
25.0
3Claude Haiku 5.5— · Cua Driver Bench on AgentEnv; Claude Code 2.1.280 with no --effort passed and its built-in tools limited to Skill and Read; trycua's cua-driver MCP tools alone; 45 turns and 360 seconds per trial
20.0
Loading Atlas data…