Tag
Microsoft open-sourced the VibeVoice speech AI framework, which supports one-shot transcription of 60-minute long audio, multi-speaker diarization and timestamp labeling, and also provides multi-role TTS synthesis capabilities. It is based on Qwen2.5 and comes with a 0.5B lightweight real-time version. It has received 24.8k stars on GitHub.
Microsoft AI announces MAI-Thinking-1, a 35B active/1T total MoE reasoning model competitive on STEM and coding tasks, developed using Ray for distributed training and orchestration.
Microsoft introduces 'average token usage' as a new metric on model release cards to measure intelligence per dollar, shifting AI competition toward efficiency and cost-effectiveness. This metric benchmarks models on both performance and the cost of achieving that intelligence.
Microsoft, Atom Computing, and EeroQ provide progress updates on their quantum computing efforts, with Microsoft demonstrating improved stability in topological qubits using new materials, while other companies share incremental advances.
Gergely Orosz recounts Kelsey Hightower's inspiring journey from DSL technician to Distinguished Engineer at Google and Microsoft, with key lessons on side hustles, explaining startups without AI, and compensation negotiation.
Microsoft, at its Build conference, announced a suite of new AI products and in-house reasoning models, signaling its intent to compete directly with OpenAI and other top AI labs after effectively separating from its long-time partner.
Microsoft released ASSERT at Build 2026, an open-source framework that converts natural language behavior specifications into executable evaluations for AI agents.
mimalloc is an open-source, high-performance, scalable memory allocator that serves as a drop-in replacement for malloc and free. Designed for modern highly concurrent applications and large memory scales, it is used in major services like Bing and integrated into projects such as NoGIL CPython and Unreal Engine.
Microsoft unveiled two new Surface devices at Build, the Surface Laptop Ultra and Surface Dev Box, both powered by Nvidia's RTX Spark chips. The Laptop Ultra features a bright mini LED display, a haptic trackpad, and strong performance including local AI model execution.
Microsoft has released Coreutils for Windows, a set of native Unix-style core tools (such as cat, cp, ls, etc.), aimed at reducing cross-platform script and development environment differences. It can be installed via WinGet and is currently in preview.
Papers With Code announced a new feature supporting closed-source evaluations, using the dense Microsoft MAI Thinking 1 tech report as an example with a special 'closed' tag.
Openclaw and Microsoft form an unexpected alliance, with Microsoft's enterprise expertise potentially supporting new developments.
Microsoft announces Intelligent Terminal 0.1, an open-source experimental fork of Windows Terminal with native AI agent integration, including an agent pane, automatic error detection, and agent management.
Microsoft is reaffirming its controversial claims about quantum computing advancements, sparking debate in the tech community.
Microsoft announced two new on-device AI models at Build 2026: Aion 1.0 Instruct, an open-weights small language model, and Aion 1.0 Plan, a 14B parameter reasoning and tool-calling model for local agentic workflows.
Microsoft releases technical details of MAI-Thinking-1 training: uses purely human data to train a base model, then trains three domain expert models, merges capabilities back into the base model via distillation, and then applies reinforcement learning to enable the model to flexibly utilize different capabilities.
Microsoft announced two new LLMs: MAI-Thinking-1 (35B reasoning model) and MAI-Code-1-Flash (5B code model), both trained on enterprise-grade, clean data without third-party distillation, with MAI-Thinking-1 claimed to be preferred over Sonnet 4.6 in blind evaluations.
Recommended reading: the MAI-Thinking-1 technical paper, which details almost all the steps to train a SOTA large language model.
Microsoft's Project Solara is a conceptual Android-based operating system designed for AI agents instead of traditional apps, showcased through two concept devices. The company plans to demo it with industry partners, though it remains non-functional at this stage.
Microsoft CEO Satya Nadella announces a shift from traditional operating systems and apps to an agent-first computing future with Project Solara, a chip-to-cloud platform for AI agents developed in partnership with Qualcomm.