Google Cloud offers Mistral AI’s Le Chat Enterprise and Mistral OCR 25.05 models.
Google Cloud is dedicated to giving consumers access to an open and adaptable AI environment so they may create solutions that are customized to meet their unique requirements. Google Cloud has announced enhanced AI solutions with Mistral AI as part of this commitment.
Google Cloud offers two distinct Mistral AI products:
- Google Cloud Marketplace’s Le Chat Enterprise
- Vertex AI’s Mistral OCR 25.05
Mistral AI Le Chat Enterprise on Google Cloud Marketplace
Le Chat Enterprise is characterized as a generative AI work assistant with a wealth of features. The Google Cloud Marketplace offers it. Connecting tools and data into a single platform is its main goal in order to increase productivity.
Le Chat Enterprise provides numerous features and functionalities on a single platform, such as:
- Custom data and tool connectors (supporting Google Drive, Sharepoint, OneDrive, Google Calendar, and Gmail at first, with more to come, including templates to create your own)
- Enterprise search
- Agent builders
- Document libraries, which let users save a private collection of commonly used documents for referencing, extracting, and analyzing from Drive, Sharepoint, and uploads.
- Personalized models
- Hybrid implementations
- Future MCP support for connecting to even more corporate systems; Auto Summary for speedy file preview and consumption; and secure data, tool connections, and libraries
The Mistral Medium 3 model from Mistral AI powers Le Chat Enterprise. It seeks to offer AI productivity on a single, highly configurable, and private platform. Self-hosted, in your public or private cloud, or as a service hosted in the Mistral cloud are just a few of the flexible deployment options that give you the freedom to select the best infrastructure without being locked in. Its privacy-first data connections and stringent ACL observance guarantee data security and protection.
From models and the platform to interfaces, users have complete control and configuration options throughout the stack. Bespoke integrations with enterprise data, unique platform/model features such as allowing user feedback loops for ongoing model self-improvement, and customizing assistants with stored memories are all examples of customization. Along with extensive audit logging and storage, it offers complete control within the security domain. Additionally, Mistral provides value delivery assistance and enhanced solutioning through its AI scientists and engineers.
A few of Le Chat Enterprise use cases:
- Creating agents: Users can create and implement a variety of agents, including no-code agents, that are aware of and synchronized with their particular context.
- Speeding up research and analysis: Makes it possible to quickly summarize long reports, extract important information from documents, and do short web searches.
- Producing actionable insights: It can produce text-to-SQL queries for financial research, automate the creation of financial reports, and transform complex data into actionable insights for sectors like finance.
- Quickening software development: Helps with code generation and review, technical documentation creation, and debugging and optimizing existing code.
- Improved content creation: Canvas lets marketers collaborate on visual assets, campaign performance analysis, and marketing copy.
The Google Cloud Marketplace lets organizations adopt Le Chat Enterprise to take advantage of Google Cloud’s scalability and security. It also simplifies procurement and connects to Google Cloud services like BigQuery and Cloud SQL.
Users can contact the Mistral AI sales team and visit the Le Chat Enterprise Google Cloud Marketplace page to begin using Mistral’s Le Chat Enterprise. Mistral AI’s announcement contains more details. You can also test out Le Chat’s features for free via Le Chat (chat.mistral.ai) or through their mobile applications.
Mistral ocr 25.05 model llm
A new OCR (Optical Character Recognition) API is called Mistral OCR 25.05. Vertex AI Model Garden has it available. Particularly noteworthy is this model’s superiority in document comprehension. It establishes a new benchmark in this field and can accurately and cognitively understand aspects of content-rich papers, including text, media, charts, tables, graphs, and equations. It extracts content in the form of ordered interleaved text and graphics from inputs such as PDFs and photographs.
How much does Mistral OCR cost?
When combined with a Retrieval Augmented Generation (RAG) system that accepts multimodal documents as input, Mistral OCR is marketed as the perfect model. Additionally, millions of users on Le Chat now employ Mistral OCR as their default model for document interpretation. The Mistral-ocr-latest API is accessible on Mistral’s developer suite, Platform, and will soon be made available on-premises and to cloud and inference partners. The price of the API is 1000 pages/$ (about twice as many pages per dollar when batch inference is used).
Mistral OCR’s highlights include:
State-of-the-art comprehension of complicated documents, which excels in understanding mathematical expressions, tables, interleaved images, and sophisticated layouts like LaTeX formatting, allowing for a deeper comprehension of rich materials like scientific articles.
Capable of parsing, comprehending, and transcribing thousands of scripts, fonts, and languages, this system is naturally multilingual and multimodal. For both global and hyperlocal enterprises, this is essential.
Excellent benchmarks: In demanding benchmark testing, this model has continuously surpassed other top OCR models. When compared to Google Document AI, Azure OCR, Gemini models, and GPT-4o, Mistral OCR 2503 exhibits the greatest scores in a number of document analysis categories, including Overall, Math, Multilingual, Scanned, and Tables accuracy. In comparison to Azure OCR, Google Doc AI, and Gemini-2.0-Flash-001, it also displays the highest Fuzzy Match in Generation score as well as the highest scores across a variety of languages. Notably, it can extract embedded pictures in addition to text, something that other LLMs examined in the given benchmark are unable to do.
The lightest and fastest in its class, processing up to 2000 pages per minute on a single node, surpassing peers in processing speed.
The use of documents as prompts for strong, clear instructions is introduced by the structured output known as “doc-as-prompt.” This makes it possible to extract particular data and format it into structured outputs, such as JSON, which can then be linked into function calls further down the line to create agents.
Selectively available for self-hosting within their own infrastructure for organizations with strict data privacy requirements for classified or sensitive information.
Mistral OCR 25.05 model example
The following are some examples of use cases for Mistral OCR 25.05:
- Digitizing scientific research: Streamlining scientific workflows by transforming papers and journals into AI-ready forms that downstream intelligence engines may access.
- Helping to digitize historical records and artifacts for preservation and improved accessibility is one way to preserve historical and cultural heritage.
- Simplifying customer support: converting manuals and documentation into indexed knowledge to increase customer satisfaction and response times.
- Preparing literature for AI in a variety of fields: In order to gain insights and boost productivity across massive document volumes, we assist businesses in converting technical literature, engineering drawings, lecture notes, presentations, regulatory filings, and more into indexed, answer-ready formats.
A complete AI platform is offered by integrating Mistral OCR 25.05 as a Model-as-a-Service (MaaS) on Vertex AI. With enterprise-grade security and compliance, it enables customers to grow with confidence and scale with fully managed infrastructure. The Vertex AI Model Garden offers a carefully chosen collection of more than 200 foundation models, including Mistral OCR 25.05, enabling users to select the best option for their unique requirements. Mistral OCR 25.05 joins additional models, like Claude Opus 4 and Claude Sonnet 4 from Anthropic, that are already accessible on Vertex AI.
Users must go to the Mistral OCR 25.05 model card in the Vertex AI Model Garden, choose “Enable,” and then follow the directions to begin building with Mistral OCR 25.05 on Vertex AI. The API is accessible on the Platform, and Mistral OCR features can also be tested out for free on Le Chat.