title: "Codestral Mamba | Mistral AI"

Mistral AI has released Codestral Mamba, a new 7.2B parameter model based on the Mamba architecture, which offers an alternative to traditional Transformers. This model provides the advantages of linear time inference and the theoretical ability to model sequences of infinite length, making it highly efficient for code generation. It is trained with advanced code and reasoning capabilities, performing on par with other state-of-the-art models. The model is available for free use under the Apache 2.0 license and can be deployed using tools like the mistral-inference SDK and TensorRT-LLM, with weights accessible on HuggingFace.