v0.15.5-rc3
📦 ollamaView on GitHub →
✨ 5 features
Summary
This release introduces two new models, GLM-OCR and Qwen3-Coder-Next, and enhances functionality with sub-agent support for `ollama launch` and VRAM-aware default context length settings.
Migration Steps
- Users relying on specific default context lengths might need to explicitly set them if the new VRAM-based defaults are not desired.
✨ New Features
- Added GLM-OCR, a multimodal OCR model for complex document understanding.
- Added Qwen3-Coder-Next, a coding-focused language model optimized for agentic coding workflows.
- Introduced sub-agent support for `ollama launch` for planning and deep research tasks.
- Ollama now defaults context lengths based on available VRAM: 4,096 (< 24 GiB), 32,768 (24-48 GiB), and 262,144 (>= 48 GiB).
- Added GLM-4.7-Flash support on Ollama's experimental MLX engine.