How to set max_tokens and chat_template_kwargs in model-management?

Before, we can set max_tokens and chat_template_kwargs(in connection config) in yaml.
Now we can only set llm config in http://localhost:8888/admin/#model-management. I can not see these config.
And when i use llm in workflow, i see the max_tokens for a llm is 2000 token.
For qwen3-397B, the enable thinking not work.

Start in the admin model-management page at http://localhost:8888/admin/#model-management and compare its available settings with the earlier YAML connection configuration. Reproduce the workflow behavior for Qwen3-397B, checking the effective max_tokens value and chat_template_kwargs; done means both settings can be configured and are honored when the model runs.

Written by the indexing model from the issue text.