Request for thinking toggle support in LM Studio
Could we have support for the thinking toggle in LM Studio, as with Qwen3.5?
Also, it will be good if other models like Gemma 4 and GPT-OSS also could have this feature. GPT-OSS does not have a thinking toggle, but I remember it has 3 configurable levels of reasoning effort.
unsloth is making a app like LMstudio soon i think
@curtis1969 Yes there is Unsloth Studio, but that is not what I am talking about. I am specifically talking about LM Studio, there is no reason to switch away from it yet, unlike Ollama.
@wqerrewetw What I did was that I copied the folder and the manifest.json and the model.yaml file from unsloth's qwen3.5 template and renamed them accordingly, replacing "qwen3.5" with "qwen3.6". It seems to work, but I get 2 separate entries of the model in LM Studio.
This model is incredible I made it 3D Jenga tower with physics at 2bit crazy
unsloth is making a app like LMstudio soon i think
We already released it btw, you can install it here: https://github.com/unslothai/unsloth
Could we have support for the thinking toggle in LM Studio, as with Qwen3.5?
Also, it will be good if other models like Gemma 4 and GPT-OSS also could have this feature. GPT-OSS does not have a thinking toggle, but I remember it has 3 configurable levels of reasoning effort.
FYI, this is something the LM Studio developers would need to implement, since they can enable it automatically for every thinking model just like how they enable it for vision models. Unfortunately, if we had to handle it ourselves, we’d need to configure each model manually, which isn’t feasible.