Mistral AI has launched Codestral Mamba, a new language model for code generation. Built on the Mamba2 architecture, it offers linear time inference and can handle infinite-length sequences. This model is designed for code productivity and has been tested with inputs up to 256,000 tokens. It outperforms other models like CodeLlama 7B and DeepSeek in benchmarking tests. Codestral Mamba is available under the Apache 2.0 license, promoting open collaboration and innovation. The Mamba2 architecture simplifies the attention mechanisms used in traditional transformer models, resulting in faster inference times and the ability to handle longer context windows. This makes Codestral Mamba particularly suitable for extensive coding tasks and local code projects. The model has 7 billion parameters and supports a 256k token context window, allowing it to manage large and complex codebases efficiently.
Mistral AI has made Codestral Mamba available for free use, modification, and distribution through platforms like HuggingFace and GitHub. Developers can deploy the model using the mistral-inference SDK or TensorRT-LLM, and support for local inference is expected to be available soon. This open-source approach encourages collaboration and innovation within the AI and coding communities.
In addition to Codestral Mamba, Mistral AI has also released a math-solving model called Mathstral, designed for math-related reasoning and scientific discovery. This model has a 32K context window and is also available under the Apache 2.0 license. Mistral AI’s commitment to open-source development and cutting-edge AI research continues to drive advancements in the field of artificial intelligence.
Our Innsights: The launch of Mistral’s Codestral Mamba highlights the transformative potential of advanced AI models in code generation. Built on the Mamba2 architecture, it offers linear time inference and handles infinite-length sequences, significantly enhancing code productivity. At InnoWave, we recognize the power of cutting-edge AI solutions. Our Data & AI offering are designed to harness the full potential of AI, driving innovation and efficiency in your business operations. With InnoWave, you can leverage state-of-the-art AI models to optimize processes, enhance decision-making, and stay ahead of the competition
Check out our Data Intelligence offer here!
Image Credits: VentureBeat made with Midjourney V6