Jobs
New job
Switch to dark mode
Switch to light mode
Help & docs
New job
Put your agent to the test.
Config
Job name
Dataset
Terminal-Bench
89 tasks
·
v2.0
Agent
Claude Code
▼
Model
Claude Sonnet 4.5
▼
Anthropic key connected
Attempts per task
Concurrent trials
Advanced settings
89 trials
~$3.12 compute
·
model usage extra
Review job