goose 搭配 MiniMax M2.5 在 OfficeQA 上以约 1/30 成本接近前沿模型
Approaching frontier performance for 1/30th the cost | goose | Your open source AI agent
goose 官方博客称，用 goose 搭配 MiniMax M2.5 在 Databricks 的 OfficeQA 基准上接近前沿模型表现，成本约为 1/30。闭源模型在准确率上占优（约 80% 对约 70%），但 MiniMax M2.5 平均每次运行 1.74 美元，Opus 4.5 为 56.53 美元。改用 goose 的 harness 后准确率比替代方案高约 10%，成本低 8 倍；在 Terminal Bench 2.0 上，goose 的 token 效率比 OpenHands 高 20 倍，成本比 OpenCode 低 40 倍以上。
推荐理由
精选AI 评分 64
原文给出 goose 搭配 MiniMax M2.5 在 OfficeQA 与 Terminal Bench 2.0 上的成本与准确率对比，可据此判断开源组合的性价比边界。
主题
正文
当前显示原文，中文全文译文尚未就绪。
Quick highlight: Sentient AGI recently shared with us some results with a challenge called "Grounded Reasoning" where teams tackled the Databricks OfficeQA benchmark using the MiniMax (2.5) open model.
Highlights from the linked article:
Using goose with MiniMax M2.5 on the OfficeQA benchmark from Databricks, we're seeing results approaching frontier model performance for 1/30th of the cost. Closed-source wins on accuracy (~80% vs ~70%), but MiniMax M2.5 ran at an average of $1.74 per run versus $56.53 for Opus 4.5. That's roughly 30× cheaper. Switching to the @goose_oss harness beat alternatives by ~10% in accuracy and was 8× cheaper than the next option. We decided to further investigate this with independent testing. On Terminal Bench 2.0, we found a similar story—goose was 20x more token efficient (and cheaper) than OpenHands, and more than 40x cheaper than OpenCode.
Check out the full post and results here: