Microsoft Build Recap in 82 seconds
Microsoft Build in San Francisco featured seven new in-house AI models, including a flagship reasoning model, a coding model, a transcription model, and a voice generation model. Microsoft also entered the AI agent space with Microsoft Scout, giving OpenAI direct access to Microsoft products and Windows management. The announcements signal a major push by Microsoft into competitive AI across multiple domains.
Summary
At Microsoft Build in San Francisco, Microsoft unveiled seven new AI models developed entirely in-house under the Microsoft AI brand. The centerpiece was a new thinking model positioned as their flagship reasoning model, designed for complex problem-solving tasks.
On the coding side, Microsoft introduced MAI Code One Flash, which was highlighted as being more accurate and more token-efficient than Claude Haiku 4.5, making it a competitive option for developers seeking cost-effective coding assistance. An ultraefficient flash variant was also announced, and despite its efficiency focus, it still ranks number two globally in image editing, narrowly behind GPT Image 2.
Microsoft also made strides in audio AI, announcing MAI Transcribe 1.5, which was claimed to be the best transcription model in the world at the time of the announcement. Complementing this, MAI Voice 2 was introduced as a speech generation model supporting 15 languages, with a live audio sample demonstrating its natural-sounding conversational output.
Finally, Microsoft entered the agentic AI space with the introduction of Microsoft Scout. This agent gives OpenAI direct integration into the broader Microsoft ecosystem, enabling it to operate across cloud, desktop, and web environments. It connects to Teams, Outlook, OneDrive, and SharePoint, with access to chats, emails, calendars, and contacts, and can directly manage Windows on behalf of users.
Key Insights
- The speaker claims MAI Code One Flash is more accurate and uses significantly fewer tokens than Claude Haiku 4.5, positioning it as a more efficient alternative for coding tasks.
- The speaker notes that Microsoft's new ultraefficient flash variant still ranks number two in image editing globally, only narrowly behind GPT Image 2, despite being optimized for efficiency.
- The speaker states that MAI Transcribe 1.5 is currently the best transcription model in the world, a bold claim that positions Microsoft above existing leaders in the transcription space.
- The speaker describes Microsoft Scout as giving OpenAI direct access to manage Windows and integrate across Microsoft's full product suite including Teams, Outlook, OneDrive, and SharePoint.
- The speaker highlights that Microsoft announced seven new models all developed in-house at Microsoft AI, signaling a strategic shift toward building proprietary models rather than solely relying on OpenAI partnerships.
Topics
Transcript
[0:00] I just got back from Microsoft Build out in San Francisco and there was a ton of announcements. They announced seven new models developed in-house at Microsoft AI. They built a new thinking model which is their new flagship reasoning model. They also introduced a new coding model MAI code one flash. It's more accurate and uses quite a bit less tokens than Claude Haiku 4.5. They also announced a new ultraefficient flash variant. and it is still ranked number [0:30] two in image editing, only very slightly being beat out by GPT image 2. They also announced a new transcription model, MAI transcribe 1.5, and it's currently the best transcription model in the world. And they announced MAI…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Matt Wolfe
AI News: GPT-5.6 and the new Super App are a Massive Leap!
OpenAI released GPT-5.6, a major leap forward in AI capabilities, alongside the new unified ChatGPT work app integrating code, browsing, and agent features. The week also saw significant model releases from xAI (Grock 4.5), Meta (Llama Spark 1.1), and research from Anthropic on AI reasoning patterns, establishing a new competitive landscape in AI development.
I Built A Monetizable Business With AI
The creator built a monetizable finance dashboard in one day using Hyper Agent from Airtable, which deploys a team of AI agents that continuously monitor news, track funding rounds and IPOs, analyze market sentiment, and generate daily briefings without manual intervention. The system demonstrates how specialized AI agents can work together autonomously to create a functional business product.
The ONLY AI Benchmark You Need!
A developer created "Buccy Bench," a humorous yet functional AI benchmark that tasks different language models with drawing Gary Busey as SVG code rather than images. The benchmark tracks model evolution over time while measuring performance metrics like cost, tokens, and execution speed.
GLM-5.2 - The Open Model That's As Good As Opus!
A comprehensive review of GLM-5.2, an open-weight Chinese AI model with a 1 million token context window, demonstrating its capabilities for coding, document analysis, and agentic workflows at significantly lower costs than frontier models like Claude Opus and GPT-4.5. The speaker tests various use cases including website building, Chrome extensions, game development, and data organization, concluding it's valuable for long, code-heavy, token-expensive tasks despite not universally outperforming closed-source alternatives.
Don't Fall For This AI Trap
The speaker emphasizes that power users distinguish themselves by knowing what NOT to automate with AI, rather than automating everything. They argue that AI works best for clear, straightforward tasks but struggles with nuanced, artistic work requiring consistency—using their failed YouTube thumbnail automation as an example.