https://mistral.ai/news/codestral-2501/

Mistral AI introduced an updated model Codestral 25.01, which improves the speed and accuracy of code generation, especially in "fill-in-the-middle" (FIM) tasks.

It really generates quickly. For some reason, the model has become worse in Java, but added % HumanEval in Python / C++ / Javascript

You can check it through Continue.dev or via the openai compatible API (codestral-latest model) - key in the console

The model in the console is not yet displayed in the limits section, most likely now is the test period.

VS code - Cline [3.1.9] & Roo Cline [3.1.6]

• Add Mistral API provider with codestral-latest model

#newllmmodel