Open-Source AI
Hugging Face, GitHub, Ollama, local models, agent frameworks
The lead
Latest first
Latest in Open-Source AI
Google’s Gemini 3.7 Flash lands in GitHub Copilot
Google's Gemini 3.7 Flash is rolling out in GitHub Copilot for Pro, Business, and Enterprise users, billed at provider list pricing via…
GitHub’s Agent Plugins 1.0 ends AI plugin duplication
GitHub and partners shipped Agent Plugins 1.0, an open standard that bundles AI skills and MCP servers into one portable plugin for…
Docker ships its own hypervisor in Desktop beta
Docker's public beta of Docker VMM swaps the third-party VM layer in Docker Desktop 4.86 for a container-tuned hypervisor, speeding startup and…
GitHub Enterprise 3.22 RC enables offline Copilot CLI use
GitHub Enterprise Server 3.22 release candidate lets air-gapped teams run Copilot CLI on-prem, with Enterprise Teams now generally available.
Docker pairs ESP32 firmware builds with AI agent sandboxes
Docker's new walkthrough builds ESP32 firmware in reproducible containers and lets AI coding agents flash real hardware from isolated Docker Sandboxes.
CORA-Diff speeds up diffusion LLM inference
A training-free method called CORA-Diff accelerates diffusion language model decoding by up to 13x on key benchmarks while preserving task accuracy.
GitHub Copilot for JetBrains gains memory and Ollama
Persistent chat memory, Ollama local models, and new enterprise controls arrive for JetBrains developers, changing how Copilot keeps project context.
Docker and partners publish AI agent security rules
Docker, Snyk and Keycard released Agent Baseline v1.0-draft, a 35-control blueprint for limiting what enterprise AI agents can reach, use and do.
Microsoft’s MAI-Code-1.1-Flash lands in GitHub Copilot
Microsoft's MAI-Code-1.1-Flash adds vision and coding gains inside GitHub Copilot while cutting list price 73% versus 1.0, reshaping low-cost AI coding.
GitHub retires MAI-Code-1-Flash on September 10
GitHub is pulling MAI-Code-1-Flash from every Copilot surface on September 10, 2026, and teams pinning the old model must switch before it…