Introducing Mistral’s Latest Innovations: Unveiling New Services and SDK for Enhanced Model Customization

French AI startup Mistral is introducing new AI model customization options, including paid plans, to let developers and enterprises fine-tune its generative models for particular use cases. Mistral has released a software development kit (SDK), Mistral-Finetune, for fine-tuning its models on workstations, servers and small datacenter nodes. For developers and companies that prefer a more managed solution, there’s Mistral’s newly launched fine-tuning services available through the company’s API. Compatible with two of Mistral’s models for now, Mistral Small and the aforementioned Mistral 7B, Mistral says that the fine-tuning services will gain support for more of its models in the coming weeks. Lastly, Mistral is debuting custom training services, currently only available to select customers, to fine-tune any Mistral model for an organization’s apps using their data.

Mistral, a revolutionary AI startup from France, is paving the way for a new level of customization in AI models. These exciting new options, including the availability of paid plans, will allow developers and enterprises to fine-tune Mistral’s generative models to suit their specific use cases.

One of the exciting new features is the self-service option. Mistral has introduced the Mistral-Finetune software development kit (SDK), which allows for fine-tuning of models on a range of systems, from workstations to small datacenter nodes.

“The SDK, which is available on Mistral’s GitHub repository, is optimized for multi-GPU setups, but can also be used with a single Nvidia A100 or H100 GPU to fine-tune smaller models, such as the Mistral 7B,” notes the company.

In fact, using Mistral-Finetune, fine-tuning a model like Mistral 7B on a dataset like UltraChat, which contains over 1.4 million dialogues with OpenAI’s ChatGPT, takes as little as half an hour with eight H100 GPUs.

For those who prefer a more managed approach, Mistral also offers a new fine-tuning service, accessible through their API. This service currently works with two of the company’s models, Mistral Small and Mistral 7B, with plans to expand to more models in the near future.

But that’s not all – Mistral also offers custom training services for select customers. This highly specialized service allows organizations to fine-tune any Mistral model for their specific apps using company data. “This approach enables the creation of highly specialized and optimized models for their specific domain,” explains Mistral in a blog post.

This news comes at an exciting time for Mistral, as they recently announced their plans to raise $600 million at a $6 billion valuation from investors such as DST, General Catalyst, and Lightspeed Venture Partners. As they face increasing competition in the rapidly-evolving generative AI space, Mistral is likely looking to grow their revenue and solidify their position as a leader in AI customization.

Since the release of their first generative model in September 2023, Mistral has continued to innovate, introducing new models such as a code-generating model and launching paid APIs. However, their user base and revenues remain undisclosed.

Avatar photo
Kira Kim

Kira Kim is a science journalist with a background in biology and a passion for environmental issues. She is known for her clear and concise writing, as well as her ability to bring complex scientific concepts to life for a general audience.

Articles: 867

Leave a Reply

Your email address will not be published. Required fields are marked *