v0.14.0-rc7
📦 ollamaView on GitHub →
✨ 5 features🐛 4 fixes🔧 1 symbols
Summary
This release introduces experimental support for image generation models via MLX, adds Anthropic API compatibility, and includes several stability improvements related to VRAM handling and error reporting.
✨ New Features
- CLI command `ollama run --experimental` now opens a new Ollama CLI featuring an agent loop and the `bash` tool.
- Added support for Anthropic API's `/v1/messages` endpoint.
- Introduced a new `REQUIRES` command in `Modelfile` to specify required Ollama version for a model.
- Ollama's app now highlights Swift source code.
- Added experimental support for image generation models powered by MLX.
🐛 Bug Fixes
- Prevented integer underflow on low VRAM systems during memory estimation for older models.
- Improved VRAM measurement accuracy for AMD iGPUs.
- An error is now returned if embeddings result in `NaN` or `-Inf`.
- Ollama's Linux install bundles now use `zst` compression.