Mistral AI, a French firm, has quickly ascended as a notable player in the artificial intelligence arena, positioning itself as a potential rival to OpenAI. While its $6 billion valuation marks it as a promising tech startup, particularly within Europe, its global market share is still developing.
Le Chat and Mobile App Launch
Mistral AI has developed Le Chat, an AI assistant, and has also released several foundational models. The launch of its chat assistant on mobile platforms generated considerable buzz, particularly in France, with French President Emmanuel Macron encouraging citizens to download Le Chat over alternatives like OpenAI’s ChatGPT.
Le Chat boasts image generation capabilities powered by Black Forest Labs Flux Ultra, enabling users to generate photorealistic images and corporate creatives. It also provides coding assistance across more than 80 languages, equipped with advanced coding capabilities that are used by developers for debugging, scripting, testing, and code optimisation.
A code interpreter enables sandboxed code execution, scientific analysis, visualisation creation, and simulation runs, making it a tool for validating algorithms and exploring data insights. Mistral plans to introduce an opt-in “Memories” feature that allows Le Chat to learn user preferences to enable personalised learning and track progress.
Mistral AI aims to put frontier AI in the hands of everyone and advocates for openness in the AI industry. The company has secured substantial funding since its inception in 2023. Le Chat reached 1 million downloads within two weeks of its mobile release, becoming the top free download on the iOS App Store in France.
The company’s founders, Arthur Mensch (CEO), Timothée Lacroix (CTO), and Guillaume Lample (CSO), have backgrounds in AI research from companies like Google’s DeepMind and Meta. Advisors Jean-Charles Samuelian-Werve, Charles Gorintin, and former digital minister Cédric O also contribute to the company.
Model Suite
Mistral AI’s suite of models includes:
- Mistral Large 2 – This serves as the primary large language model.
- Pixtral Large – Unveiled in 2024, this addition to the Pixtral family handles multimodal tasks.
- Codestral – This is a generative AI model that specialises in code-related tasks.
- Les Ministraux – This is a family of models optimised for use on edge devices like smartphones.
- Mistral Saba – This is focused on Arabic language processing.
Monetisation Strategies and Partnerships
Mistral AI distinguishes between its premium models, which are not open source, and its free models, which are available under the Apache 2.0 license. Free models include research models like Mistral NeMo, which was built with Nvidia and open-sourced in July 2024.
To generate revenue, Mistral AI offers paid tiers for Le Chat, with the Pro plan priced at $14.99 per month. It also monetises its premium models through APIs with usage-based pricing for businesses. Strategic partnerships also contribute to revenue.
A deal with Microsoft includes distribution of its AI models through Azure and a €15 million investment. Mistral AI has also partnered with Agence France-Presse (AFP) to allow Le Chat to query AFP’s text archive dating back to 1983. Further partnerships include France’s army and job agency, German defence tech startup Helsing, IBM, and Orange.
Orange and Mistral AI are collaborating to improve AI capabilities across Europe. The partnership includes research to design network infrastructure for AI, assess the impact of AI on telecommunications networks, and define technology roadmaps. Orange will integrate Mistral’s applications, such as Le Chat Pro and Codestral, into its service offerings.
Meanwhile, Mistral is partnering with Helsing, a European defence company, on “Vision-Language-Action models” for edge devices.
What the Future Looks Like
As of February 2025, Mistral AI has raised approximately €1 billion (about $1.04 billion) in capital. CEO Arthur Mensch has stated that Mistral is “not for sale” and intends to pursue an IPO.
Mensch believes the AI industry will shift from focusing on individual models to systems that integrate models with contextual business data. He envisions AI systems enhancing the capabilities of large language models by linking them to relevant data, improving their ability to provide customised solutions.
Mistral is planning to invest billions in an AI data centre in Essonne to train more efficient systems.