Re:Cursor Composer: Building a fast frontier model with RL
by chaidhat on Wednesday October 29, 2025 @01:44PM
(#45750422)
Which model did you distill it from? Great work! PS getting a few scenarios where it doesn't follow rules as well as sonnet 4.5
[ Parent
| on HN ]
Re:Cursor Composer: Building a fast frontier model with RL
by srush on Wednesday October 29, 2025 @01:48PM
(#45750484)
The blog talks about the training process. Specifically we trained with RL post-training on coding examples.
Re:Cursor Composer: Building a fast frontier model with RL
by chis on Wednesday October 29, 2025 @01:58PM
(#45750637)
Makes sense, but what model was used for the base? Is it some open-source model, and you're not at liberty to disclose?
Re:Cursor Composer: Building a fast frontier model with RL
by W0WL0LXD on Friday October 31, 2025 @02:46PM
(#45775316)
not a Cursor employee but still a researcher, it’s Zhipu/Z.ai GLM-4.6/4.5. there’s traces of Chinese in the reasoning output + its the only model that would make sense to do this with RL, and is a model that already delivers near SOTA performance + is open-source/open-weight.
Cursor Composer and Windsurf SWE 1.5 are both finetuned versions of GLM.