Editorial investigation
Raw coding view, stripped back
Inspect the independent evaluation configuration and protocol on the model page.
Open top coding modelLoading the current published AI model ranking…
Coding ranks the current published roster by comparable Terminal-Bench 4.0 results: mini-swe-agent, three repeats, average pass@1.
Buyer-facing table
View explanation
Why the Coding view changes
Coding ranks the published roster by Terminal-Bench 4.0, using mini-swe-agent, three repeats and average pass@1.
Editorial investigation
Inspect the independent evaluation configuration and protocol on the model page.
Open top coding model"Worth trialling for coding and extended research, with hosted access and a separately billed API."
Kimi K3.1 Pro
Moonshot AI
Practical computer tasks done at a command line, such as setting up software, fixing broken code or processing files. Shows how well the model can work as a hands-on coding or IT assistant.
How to read it: Percentage of tasks completed successfully. Higher is better.
Different Terminal-Bench versions use different task sets, so compare scores only within the same version.
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0
Terminal-Bench 4.0