News
Updates, product notes and practical coverage for AI platforms and operators.
Mistral has released Timestral, a four-model family for probabilistic multivariate time-series forecasting. The 5M-to-209M-parameter models support long contexts and covariates, can run on CPUs, and are available for zero-shot use or adaptation through Mistral Forge. Mistral reports leading public benchmark results but says a fuller technical report is still forthcoming.
Mistral has opened Managed Deployments in public preview, allowing paid AI Studio customers to run workflow workers from public or private GitHub repositories without operating infrastructure. Mistral Cloud handles builds, scaling and lifecycle management, with service-account authentication, optional blocked network egress, API controls and worker logs inside Studio.
Mistral Large 4 is now in public preview as a one-trillion-parameter, natively multimodal model with 49 billion active parameters. Mistral positions it for coding, scientific computing, enterprise knowledge work and agentic tasks, with open weights due later this month and API access already available through Mistral Studio.
Mistral has opened a Munich hub combining physics-AI research with applied engineering for German industry. The site incorporates expertise gained through its Emmi AI acquisition and supports work with BMW, Siemens Energy and the Technical University of Munich. It forms part of Mistral’s broader European sovereignty and compute strategy.
Mistral and Mozilla have announced a partnership bringing Mistral models to Firefox Smart Window. The beta rollout covers France and North America, with further markets planned. Privacy controls and model choice are central to the offer, but local memory storage should not be confused with entirely on-device AI processing.
Mistral and Cloudera have announced an integration covering private inference and customised models within customer-controlled environments. The partnership targets enterprises that want AI close to governed data. Deployment details, operating costs and the timing of individual capabilities still need careful assessment before the announcement becomes a production adoption plan.
Mistral has raised €3 billion in a Series D led by Samsung Electronics, valuing the company above €21 billion. The company says the capital will expand frontier research, training capacity, infrastructure and international operations around its open-weight, sovereign AI strategy.
Microsoft and Mistral have expanded their partnership through a multibillion-dollar commitment to European AI infrastructure. The agreement also brings Mistral Medium 3.5 and OCR 4 into Microsoft products, with deployment options spanning public cloud, Azure Local and disconnected environments.
Mistral Studio now gives organisations a central system for managing, testing and governing AI prompts and skills. New features include immutable versions, ownership records, rollback, classification labels, audit logs and production traceability.