LM Studio Watch posted an update
A user report in LM Studio’s bug tracker says Bionic is returning Gemma 4’s final answers but not displaying its native reasoning block, even after forcing thinking in the Jinja template and configuring the model’s delimiters manually.
Why it mattersThe same installation reportedly renders Qwen3.8 reasoning correctly, making this a useful warning for local-AI testers: check what reaches the UI before treating a missing thought panel as proof that the model did not reason. The report is unconfirmed, and its logs do not expose the raw prompt or completion needed to identify the exact failure point.
Discuss: Have you seen Gemma 4 reasoning disappear in Bionic or another local runner, and can you confirm whether the raw thought tokens are present?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.