Fetching from the wire…
Models2026-08-15 · source-backed
The update adds native image input and faster token streaming alongside gains in instruction following and tool use. It arrives with Polaris becoming the default model on every Copilot seat and Visual Studio 18.9 exposing low/medium/high thinking-effort controls. RuntimeWire Microsoft iterating its own model rather than reselling frontier vendors is the part to track, and the cost cut is what makes first-party viable. Same logic as Wix and Base 1.
Each link below shares sources, entities, or timing with this story.
On June 18 GitHub made its purpose-built small coding model available across the Copilot app, CLI, Chat, Visual Studio, JetBrains, Eclipse, GitHub Mobile, and Xcode, in Free through Max plans. (GitHub) A fast, cheap first-party model pushed this broadly signals Microsoft routi...
MAI-Code-1-Flash, a 5B-parameter coding model, is in GitHub Copilot and VS Code, and Microsoft says it beats Claude Haiku 4.5 across core coding benchmarks, +16 points on SWE-Bench Pro at 51.2% versus 35.2%, using up to 60% fewer tokens. MAI-Thinking-1, a 35B-active MoE with a...
Microsoft unveiled MAI-Code-1-Flash at Build, its first model that turns written descriptions into source code for apps and websites, explicitly aimed at lowering developer costs and cutting its dependence on OpenAI (CNBC). Same week, Google pushed Gemini 3.5 Flash to general...
The August 10 weekly release rolls Moonshot's 2.8T-parameter, 896-expert open-weight model out across all tiers, and Microsoft's own MAI-Code-1.1-Flash started rolling out with native image understanding (GitHub Changelog). Copilot is now the fastest legitimate route to K3 for...
Expressive TTS with voice cloning across 15 languages (Product Hunt), alongside MAI-Code-1-Flash. Microsoft is methodically building out an in-house model stack across code, voice, and image to depend on OpenAI less across every modality, not just one. The "rent the model" era...
Released July 27, a sparse-MoE security fine-tune of MAI-Code-1-Flash with a 256K context. Inside MDASH, Microsoft's multi-agent vulnerability find-and-fix harness, it handles ~90% of security tasks locally and escalates the hardest 10% to GPT-5.4, costing 50% less than the pr...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.