StepFun Open-Sources Step Code: Self-Tests Match DeepSeek Harness, With Lower Token Consumption

Beating AI News Flash: StepFun has open-sourced its AI coding tool Step Code v0.1.0 under the MIT license. It runs directly in the terminal, can read and write code, run tests, and execute development tasks, and also supports MCP, Agent Skills, and Claude Code plugins. The built-in StepPage can publish a local static website online with a single command.

Step Code focuses on using fewer tokens. In StepFun's own Terminal-Bench 2.1 comparison, it completed 72/89 tasks with a score of 80.9%, tying with DeepSeek Harness in the chart for the highest score, but with lower token usage. StepFun's self-built Multi-Frame long-task test has a total of 150 questions. Step Code completed 110 of them, scoring 73.3%, with an average of 5.09 million tokens consumed per question, achieving both the highest pass rate and the lowest token usage among the 6 harnesses compared.

However, StepFun did not disclose which underlying model was specifically used in this round of comparison, nor did it break down how these tokens were saved. Officially, it only mentioned that Step Code and the Step series models were tuned together for token consumption. From the source code, it is quite "stingy with tokens" throughout the harness: file reads and command outputs are actively length-limited, searches try to return only the needed parts, long conversations compress context by default, and the system prompt also requires that if a tool can be called directly, an extra Agent should not be opened. It even has a more aggressive context trimming mechanism that can delete historical tool outputs, duplicate results, and old reasoning processes, though this feature is disabled by default.