goose benchmarks five open-source coding harnesses: similar pass rates, up to 75% difference in per-task cost
Original title:Same Brain, Different Bill
AI summary
goose officially ran 310 scoring trials each for five open-source harnesses—goose, Codex CLI, OpenCode, Hermes Agent and pi—using the same model gpt-6.1-sol (high reasoning effort) and 62 tasks from Terminal-Bench 4.0.
Key Takeaways
- goose officially ran 310 scoring trials each for five open-source harnesses—goose
- Codex CLI,OpenCode,Hermes Agent and pi—using the same model gpt-6.
- 1-sol (high reasoning effort) and 62 tasks from Terminal-Bench 4.
Quick Facts
- Published
- Updated
- Source
- goose 官方博客
- Category
- Papers & Research
- AI score
- 64
同一模型换不同 harness 通过率相差不大,成本却差出 75%,这组 310 次试验的对照数据可帮读者把选型注意力从模型转向 harness。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: goose benchmarks five open-source coding harnesses: similar pass rates, up to 75% difference in per-task cost. October 28, 2026. https://new.nat6.net/en/news/fukvhnr4747eqhe0yjrn9d8nc/