Ollama v0.32.0 adds Qwen 3.5 support and agent UI
New pre-release brings Qwen 3.5 parser, agent warnings, and UI improvements.
Ollama has released version 0.32.0 as a pre-release, bringing several notable updates for local LLM enthusiasts. The primary change is the addition of a dedicated parser and renderer for Qwen 3.5 and Qwen 3.5 Next models, contributed by jessegross. This ensures that these newer Chinese-developed models are properly handled within Ollama's ecosystem, enabling users to run them locally with correct tokenization and output formatting.
Alongside model support, the release includes improvements to the agent system. A warning will now appear when users attempt to launch older agent models, helping avoid compatibility issues. Additionally, a new command-line agent UI has been implemented by ParthSareen, providing a more intuitive interface for interacting with AI agents directly from the terminal. While still a pre-release, v0.32.0 signals Ollama's ongoing commitment to supporting cutting-edge open-source models and enhancing the developer workflow.
- Adds qwen3.5 parser and renderer for Qwen 3.5/Next models
- Warns users before running outdated agent models to prevent issues
- Introduces a new command-line agent UI for easier interaction
Why It Matters
Expands Ollama's model compatibility and improves agent usability for local AI development.