French AI startup Mistral has unveiled its latest AI model, Pixtral 12B, marking its first multimodal model capable of handling both text and images. With a size of approximately 24GB and 12 billion parameters, Pixtral 12B is designed to perform tasks such as image captioning and object counting, showcasing capabilities comparable to popular models like Anthropic’s Claude and OpenAI’s GPT-4.
Built on Mistral’s text-based Nemo 12B model, Pixtral 12B allows users to input images through URLs or via base64 encoding. The model can be downloaded and customized under the Apache 2.0 license, providing open access without restrictions, a move confirmed by Mistral in an email.
Although no public demos are available yet, Mistral plans to introduce Pixtral 12B for testing on its Le Chat and Le Platforme chatbot platforms soon. However, the specific image data used to train the model remains unclear, potentially raising questions regarding copyrighted material and AI model training practices.
Pixtral 12B’s release follows Mistral’s recent $645 million funding round, led by General Catalyst, which values the company at $6 billion. Positioned as Europe’s competitor to OpenAI, Mistral has gained attention for offering open models and consulting services, with a minority ownership stake held by Microsoft.
Future plans for Mistral
Mistral, the French AI startup, has ambitious plans for the future as it continues to position itself as a leading player in the European AI space. Some of the key future goals and strategies for Mistral include:
1. Expanding Multimodal Capabilities
With the release of Pixtral 12B, Mistral has entered the multimodal AI space, combining text and image processing. The company is likely to continue advancing this technology, potentially adding video, audio, and other modalities to future models, similar to competitors like OpenAI and Anthropic.
2. Scaling AI Model Offerings
Mistral’s roadmap likely includes scaling up the size and sophistication of its models. With Pixtral 12B already housing 12 billion parameters, future models could increase in size and complexity, focusing on specific domains like healthcare, finance, or manufacturing to cater to industry-specific use cases.
3. Developing Proprietary AI Tools and Platforms
Mistral is expected to enhance its platforms like Le Chat and Le Platforme, where users can interact with and test their models. The company may introduce API services, advanced fine-tuning tools, and customizable AI solutions to meet enterprise demands for personalized AI models.
4. Strengthening Corporate Partnerships
Mistral’s strategy involves releasing open models while offering managed services and consulting to large corporations. The company will likely expand its enterprise partnerships, particularly with firms that need tailored AI models for automation, data analysis, and predictive analytics. Microsoft, a minority stakeholder, may also play a significant role in Mistral’s growth through deeper collaborations.
5. Pushing for AI Ethics and Regulatory Leadership
Based in Europe, where AI regulations are becoming stricter, Mistral will probably focus on ensuring its models comply with upcoming EU standards, such as the AI Act. This could also involve playing a part in shaping industry guidelines on ethical AI, particularly regarding data privacy, transparency, and responsible AI use.
6. Exploring New Funding Rounds and Market Valuation Growth
Following its successful $645 million funding round, Mistral may pursue additional capital to fuel its R&D efforts. Its goal could be to increase its valuation beyond the current $6 billion, possibly making the company a key player in any potential European AI IPO or further merger and acquisition activity.
7. Challenging US AI Giants
As Europe’s answer to OpenAI, Mistral will continue to position itself as a European alternative to the dominant U.S. AI firms. It may focus on gaining market share in enterprise AI and building models with language diversity and localized solutions for non-English-speaking markets, further distinguishing itself from U.S. competitors.
By continuing its focus on innovation, open-access AI models, and corporate engagement, Mistral aims to cement itself as a global leader in AI. Its next moves will be crucial in shaping the company’s role in the competitive AI landscape.