A post reports that MiMo Pro's DeepSWE score rose from 58.41 to 72.57 after reinforcement-learning runs, nearing comparative results cited in the post.
Published on September 21, 2026, the post reports that MiMo Pro's score on the DeepSWE benchmark rose from 58.41 to 72.57 after reinforcement-learning (RL) runs. It compares the updated result with a score of 74, attributed to Astra, Gemini 3.8 Flash, and Opus 5.
The figures provide a reported comparison for evaluating RL runs on DeepSWE; they are not, by themselves, an independent verification and do not provide methodological details. Consult the original post to check its context and cited results. If using AI to study or apply this material, avoid entering confidential organizational data and verify claims against the source.