Ollama releases v0.17.8, which continues to patch and enhance the native model running experience, with updates focusing on GLM tool call parse fixes, Gemma 3 vision model support, and Gemini CLI compatibility patches. For users who use Ollama in local inference, agent toolchains, and multi-model development environments, this version is not an exaggeration, but each one is close to everyday use.
According to the release note, this time it fixes the parsing problem caused by unclosed parameter tags in GLM tool calls, and also adds Gemma 3's support for image input scenarios, while allowing the Gemini CLI to run more smoothly in Ollama's workflow. In addition, the release also includes a number of community contributions, indicating that the project is still rapidly iterating on the model compatibility layer and tool call details.
The competition in the local model ecosystem increasingly depends on "whether it can run steadily". The value of updates like Ollama is not to introduce a new concept, but to continue to bridge the gaps between different models, different CLIs, and different calling methods, allowing developers to spend more time on the application itself rather than fixing the environment.
FAQs
Q: What are the core changes in v0.17.8?
A: GLM tool call parse fixes, Gemma 3 vision support, and Gemini CLI compatibility enhancements.
Q: Why is parsing fix worth paying attention to?
A: Because once the tool call parses an exception, it will directly affect the execution of the proxy and automated processes.
Q: What does Gemma 3 visual support mean?
A: This means that Ollama's local running coverage of multimodal models continues to expand.
Q: Who are the changes to the Gemini CLI more important for?
A: It is more important for developers who want to combine local model running with development toolchains.
Q: What is the overall direction of this update?
A: This reflects Ollama's continuous enhancement of local model compatibility and daily stability.
