Intel / post

StepFun Open-Sources Step Code, a CLI Coding Agent Benchmarked on Terminal-Bench

Source: x.com
Date: Published Sep 22, 2026

Context

StepFun released Step Code v0.1.0, a command-line coding that, in its words, handles the full development loop from one CLI: reading and editing code, running tests and shipping. The code is open source under the MIT license.

StepFun reports 80.9% on Terminal-Bench 2.1 and 73.3% on Multi-Frame, which it describes as its own 150-task long-horizon . These are StepFun's figures from its announcement, and the post does not say how the evaluations were run, so they are claims to check rather than an independent comparison with other coding agents.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.

Read the source x.com