Ollama 0.24 ships with support for the Codex App, a desktop environment for parallel coding threads with worktree support, a built-in browser for annotating local sites, and an in-app code review mode. It also adds a reworked MLX sampler for better generation quality on Apple Silicon.

Qwen 3.7 Max from Alibaba is now accessible through Vercel AI Gateway, bringing agent-focused capabilities including frontend prototyping, multi-file engineering, and office workflow automation to builders on the platform.

OpenAI is preparing to confidentially file its IPO prospectus as soon as Friday, working with Goldman Sachs and Morgan Stanley. The move signals the AI company is on track for a public debut as early as Q4 2026.

GitHub has disclosed an active investigation into unauthorized access to its own internal repositories. Affected customers will be notified through established incident response channels.

Grok Build 0.1, xAI's beta coding model built for agentic workflows, is now accessible via Vercel AI Gateway. It powers the Grok Build CLI and is currently in early access with no configurable reasoning settings.

Waiting for a smarter model is the wrong lever. Reuben Brooks argues that structural verification gates, baked into the code substrate, catch what prompt rules and agent instructions always miss.

OpenCompass 0.5.2 ships support for 14 new benchmarks spanning science, math, and instruction-following, plus new model and API integrations. If you evaluate LLMs in production, this release expands what you can measure out of the box.