Ollama v0.24 launches Codex App with local & cloud models
Code, browse, and annotate inside Ollama's desktop app with any local or cloud model.
Ollama's latest release, v0.24.0, brings the OpenAI Codex App directly into the Ollama ecosystem. Users can now launch a full-featured coding assistant inside the desktop app that supports browsing, local server spinning, and inline annotation — all powered by any Ollama model, whether running locally on your machine or via an Ollama Cloud subscription. This enables developers to choose the best model for each task, from lightweight local options to cutting-edge cloud models.
For heavy coding and agentic tasks, Ollama recommends the kimi-k2.6 (with vision support) and glm-5.1 models. For pure local use without a cloud subscription, the top picks include nemotron-3-super, gemma4:31b, and qwen3.6. The Codex App also includes a review mode for leaving comments and iterating without leaving the workspace. Additionally, the MLX sampler has been reworked to improve generation quality on Apple Silicon Macs.
- OpenAI Codex App now integrated into Ollama, supporting local and cloud models
- Built-in browser and review mode for annotating code and iterating inline
- Recommended local models: nemotron-3-super, gemma4:31b, qwen3.6; MLX sampler improved for Apple Silicon
Why It Matters
Gives developers flexible, high-quality AI coding assistance using any Ollama model, local or cloud.