Correct way to access the logit_bias feature

It seems the OpenAI API supports the attribute .logit_bias, which allows us to encourage or discorage / ban certain words: https://help.openai.com/en/articles/5247780-using-logit-bias-to-alter-token-probability-with-the-openai-api and the lmstudio API also supports it:
https://lmstudio.ai/docs/app/api/endpoints/openai

However, when I try to use it, it seems to have no effect.
For instance, the word "time" for Qwen3-14b is token no. 1678:

token = a.tokenize("time")
print(token)
[1678]

But if I try to pass logit_bias to Qwen 3, it ignores logit_bias completely:

self.model.act(
chat_object,
tool_list, # Pass the actual tools list
config={
"temp": 0.2,
"maxTokens": MAX_HISTORY_LENGTH,
"logit_bias": {"1678":-100},
},
on_prediction_fragment=self._stream_response,
)

[USER (10:08 AM)]: Complete this sentence: "Once upon a..."

[AI (10:08 AM)]: Once upon a time, there lived a curious little girl who loved exploring the enchanted forest behind her cottage.

Is this the right way to access logit_bias? Or is it a specific model issue?

Start at the self.model.act call and its config handling, using the Qwen3-14b reproduction with token 1678 as the test case. Compare the supplied logit_bias shape with the LM Studio and OpenAI API behavior, then establish whether the SDK applies it or whether the limitation is model-specific; done means the expected behavior is reproduced or clearly documented.