Introduction
The month of June 2026 marks an important milestone with the new updates announced by Microsoft regarding Copilot and artificial intelligence. This article provides a summary of key elements, ranging from new artificial intelligence models to enhanced features, while highlighting best practices for leveraging these tools in a professional environment.
For more details on these announcements, consult the complete Microsoft Build video this year on YouTube.
Good to Know
The following sections detail the main updates by theme, from advanced AI models to extended Microsoft Copilot integrations, allowing IT teams to optimize their workflows.
New AI Models Available on Azure
Microsoft has deployed several new artificial intelligence models on Azure, strengthening its ecosystem with advanced capabilities:
Claude and Mythos 5 Models
-
Claude Mythos 5: Designed to meet the most demanding needs in code development, document synthesis, and complex workflows. This model integrates:
- Retention of prompts and responses over 30 days for quality and security purposes.
- Usage restrictions to prevent misuse in fields such as biology or chemistry.

Attention
Mythos 5 models are strictly reserved for certain scenarios, such as cybersecurity and projects validated under the Glass Wing program.
- Gemini 2.7: Available via GitHub Copilot, this is an affordable option specially designed for code development.
Microsoft AI Models (MAI)
Microsoft is also enriching its AI portfolio with the MAI series for various use cases:
- MAI Thinking 1: Medium-sized model specialized in dialogue and reasoning.
- MAI Image 2.5: Image generator with advanced features such as image-to-image editing.
- MAI Transcribe 1.5: Multilingual transcription tool, equipped with features like speaker diarization and voice cloning.
1{2 "feature": "Voice cloning",3 "useCases": ["Multilingual conferences", "Audio archiving"]4}Tip
Use the "flash" variants of MAI models for applications where energy efficiency and costs are critical.
Work IQ API and Cost Optimizations
The new features of the Work IQ API revolutionize how AI agents interact with organizational data and the internet. Work IQ enables:
- Context Access: Returns only relevant data to optimize token costs.
- Operational Tools: Capabilities to interact with emails, calendars, and workspaces.
- Usage-Based Billing: Different cost levels are applied depending on query complexity.
Configure Copilot Credits
Access the Azure Copilot dashboard and set consumption limits to control costs associated with the Work IQ API.
Enable Integrations
Connect the Work IQ API with your internal data by configuring knowledge sources (SQL, Fabric IQ, etc.).
Good to Know
Integration of Work IQ with global Bing index via Web IQ allows combining enterprise context and online information for optimal results.
Improvements in Copilot Studio and Agent Management
Microsoft has also revised Copilot Studio to facilitate the deployment of AI agents within the Microsoft 365 ecosystem:
Custom Agents and Security
- Advanced Evaluations: Use of LLM models to evaluate AI agent performance on complex interactions.
- Continuous Optimization: Integration of automated adjustments for ongoing improvement through Foundry.
Main Updates
- Multi-Step Agents: Design of agents capable of executing complex workflows, combining AI and deterministic logic.
- Enhanced Connectivity: Direct publication of agents on Microsoft Teams and Copilot.

Important
Advanced agent security features will soon be available exclusively through the Microsoft Agent 365 (A365) license. Plan this transition in your roadmap.
Conclusion
Recent announcements around Microsoft 365 Copilot show clear evolution, with emphasis on cost optimization and integration of models into professional workflows. These tools offer enterprises an unparalleled opportunity to improve their productivity and push the boundaries of artificial intelligence in their daily operations.
To discover other tips and practical tutorials, explore our Copilot section.



