H Company released Holo4-27B and Holo4-35B-A3B, multimodal agent models that can work through graphical interfaces, code, MCP and APIs. Public checkpoints include BF16, FP8, NVFP4 and GGUF variants, while a 2.74 GB trajectory dataset exposes 7,366 benchmark runs with steps, tool results and screenshots.
Why it matters: The release makes both usable computer-use checkpoints and the underlying evaluation traces available for inspection. Teams can test one model across desktop, web, mobile, code and tool-call workflows instead of trusting aggregate benchmark scores alone.
Pollen Robotics released Grabette, a handheld gripper, camera and IMU system that records manipulation demonstrations without a robot. Its browser pipeline recovers trajectories and publishes synchronized episodes as standard LeRobot datasets.
Why it matters: Robot policy progress is constrained by varied real-world demonstrations. An open recorder can lower data-collection cost and separate capture from robot ownership.
Mistral released OCR 3 through its API and Studio for extracting text, tables, forms, handwriting, and layout from documents and images.
Why it matters: Document AI was shifting from plain text recognition to structure-preserving extraction that can feed searchable archives, workflows, and agents.
Mistral released Mistral Large 3, a 675B-parameter mixture-of-experts model with 41B active parameters, plus 14B, 8B, and 3B Ministral 3 models, all under Apache 2.0.
Why it matters: One permissively licensed family covered frontier-scale serving, enterprise customization, laptops, and edge devices, reducing the need to adopt unrelated model stacks for each tier.
Mistral introduced Memories in beta for carrying context across conversations, with export and import controls, alongside built-in and custom MCP connectors for bringing external tools and information into Le Chat.
Why it matters: Persistent context and connectors were turning assistants into personal work surfaces while raising new questions about recall, consent and data boundaries.
Mistral released Voxtral Small 24B and Voxtral Mini 3B under Apache 2.0, covering transcription, multilingual understanding, audio question answering, and function-oriented speech workflows.
Why it matters: Speech systems became easier to inspect, customize, and deploy privately, reducing dependence on closed transcription APIs for multilingual and sensitive audio.
Mistral introduced Magistral Small as a downloadable 24B reasoning model and Magistral Medium through its hosted platform, emphasizing multilingual, domain-specific, and traceable step-by-step reasoning.
Why it matters: It widened access to reasoning beyond U.S. and Chinese labs and provided a downloadable model for teams needing language coverage or local control.
Mistral and All Hands AI released Devstral Small under Apache 2.0 for resolving repository-scale software issues through agent harnesses such as OpenHands. Mistral reported 46.8% on SWE-bench Verified and said the checkpoint could run on a 32GB Mac or single RTX 4090.
Why it matters: Repository-level coding agents gained a locally deployable model option for private codebases and controlled experiments.
Mistral released Medium 3 through its platform and Amazon SageMaker, positioning the model for coding, multimodal understanding and enterprise customization through private deployment, continued pretraining and fine-tuning.
Why it matters: The release strengthened the middle tier between small local models and expensive frontier systems, where many enterprise workloads can be served more economically.
Mistral released Small 3.1 under Apache 2.0 with image understanding, a 128k context window, function calling, and local deployment guidance. It retained the 24B size of January's Small 3.
Why it matters: A relatively compact, permissively licensed model could cover document images, local assistants, and tool-using applications without a separate vision stack.
Mistral launched an OCR API that extracted ordered text and images from PDFs, including tables, equations and complex layouts, for use in document workflows and retrieval systems.
Why it matters: Reliable document extraction is a prerequisite for grounded assistants because layout, tables and figures often carry meaning that plain text conversion loses.
Mistral released a 24B-parameter model under Apache 2.0, positioned for fast conversational, function-calling, and domain-specific workloads that could run on a single high-memory GPU or local workstation.
Why it matters: It strengthened the middle tier between tiny edge models and very large frontier systems, where privacy, latency, and predictable serving cost often matter more than maximum benchmark score.
Mistral released Codestral 25.01 through its API and IDE partners, with a 256,000-token context window, fill-in-the-middle support and options for deployment inside an enterprise VPC or on premises.
Why it matters: Coding models were becoming high-frequency infrastructure for inline completion and repository work, not merely assistants used in a separate chat window.