Ollama v0.32.8 brings Muse Glimmer to all platforms
Muse Glimmer now powers Claude Code, Codex, Pi, and more locally on your machine
Ollama released v0.32.8, bringing the Muse Glimmer model to all platforms. This update adds support for NVIDIA, AMD, and additional hardware, expanding beyond the Apple Silicon MLX engine that already delivered state-of-the-art performance with DFlash and image input support in v0.32.7. Muse Glimmer is designed to power coding agent applications such as Claude Code, Codex, and Pi, as well as long-running personal assistants like OpenClaw and Hermes.
To get started, users can simply run `ollama run muse-glimmer` to download and execute the model locally. For integration with Claude Code, the command `ollama launch claude --model muse-glimmer` sets up the coding agent. Lighter-weight coding agents can be launched with `ollama launch pi --model muse-glimmer`, while personal assistant frameworks use `ollama launch openclaw` or `ollama launch hermes`. This release positions Ollama as a versatile local runtime for a growing ecosystem of AI agents.
- Muse Glimmer model now supported on NVIDIA, AMD, and other platforms
- Works with coding agents: Claude Code, Codex, Pi, plus assistants OpenClaw and Hermes
- Apple Silicon users get MLX engine optimization with DFlash and image input support
Why It Matters
Local AI agents become more accessible across hardware, enabling private and cost-effective coding and personal assistant workflows.