Interested in regular Model News & Updates? Subscribe to the The Model Mondays Newsletter - read the first edition for context.
Model Mondays is a weekly series that combines a 30-minute livestream (Monday at 1:30pm ET) with a 30-minute AMA (Fridays at 1:30pm ET) to build your AI model IQ.
- 5-min Highlights - recaps the top 5 news items from last week
- 15-min Spotlights - a deep-dive with a subject matter expert
- 30-min AMAs - open-ended discussion, Q&A with the expert
This week we put the spotlight on Advanced Reasoning with subject matter expert Marlene Mhangami - and covered the top 5 model-related news from Azure AI Foundry - in our #ModelMondays livestream. Browse the slides below - then join us Friday for Q&A with Marlene:
👉🏽👉🏽 Register for the AMA on our Azure AI Foundry Discord
👉🏽👉🏽 Post Your Questions on our Discussion Forum
Want to get a sense for what Reasoning models are, what advanced reasoning covers? Read on.
What are Reasoning models?
Reasoning models are a new category of Large Language Models that are trained (with large-scale reinforcement learning) to think deeply before they respond. These reasoning abilities are achieved by a combination of techniques including chain-of-thought, self-consistency, and deliberative alignment.
Reasoning models are designed to tackle hard problems involving logic, stratgegy, complex reasoning and multi-step planning. They are particularly effective in STEM fields (math, science, coding) where they can outperform popular general-purpose models in competitive benchmarks.
Want an introduction to reasoning models? Watch this episode from Season 1 where Jennifer Marsman covers the OpenAI o-series of reasoning models.
Advanced Reasoning Scenario
While the previous talk gives you a sense for what reasoning models are with a real-world use case, this advanced reasoning episode takes it a step further with an end-to-end sample and workshop that teaches you how to build a Deep Researcher - a research assistant that can conduct comprehensive web research, analyze & synthesize the information, and present its findings in a report.
The solution uses the DeepSeek-R1 reasoning model with the Tabily web search API, and the LangGraph framework to perform iterative cycles of research before delivering the final result. Watch the livestream of the episode below:
Reasoning Lab For Beginners
Want to get hands-on experience with reasoning models and learn how they differ from general-purpose models? Try this self-guided lab from Microsoft Build 2025 which explores the OpenAI o-series of reasoning models with hands-on exercises!
👉🏽👉🏽 Explore The Lab on GitHub
👉🏽👉🏽 Browse The Presentation on SpeakerDeck
The storyboard summarizes what the lab covers.
Build Your Model IQ!
Model Mondays Season 2 is currently scheduled to go from June to September, covering the 12 key topics shown below.
👉🏽👉🏽 Register for Upcoming Livestreams to get reminders
👉🏽👉🏽 Register for Upcoming AMAs to get reminders
This Week In Model Mondays
Want an easy way to catch up on all the news? Check out This Week in Model Mondays - a recap written by Sharda Kaur. This first post is now live
Want to get weekly updates and recaps on what was covered on Model Mondays - and a sneak peek at what's coming next? Follow Sharda Kaur - a Gold Microsoft Learn Student Ambassador - as she recaps each week with a blog post on Friday!
Top comments (2)
Pretty cool seeing this all broken down straight up, makes me wanna show up and actually try some of that hands-on lab stuff.
Check out aka.ms/discuss/build25-lab333 - it was the intro-level lab we ran at Mirosoft Build recently and it is a good exercise to try out to understand reasoning models with Azure Open AI o-series
You can then try out the advanced one that Marlene shows - which uses DeepSeek-R1
Some comments may only be visible to logged-in visitors. Sign in to view all comments.