Models
What GLM 5.3 Flash Uncensored is
GLM 5.3 Flash Uncensored is the default model for chat, text, and code. Its id is z-ai/glm-5.3-flash-uncensored. An empty model field routes to it.
This is not a Hugging Face model card and not a benchmark dump. Here it is the daily chat: text, code, and when needed images, search, and reasoning. The key comes from @uncensobot. Base URL is https://uncensored.dog/v1.
Who it is for
Everyday work: drafts, edits, short code, a reply without the “I can’t assist with that” template. If you need a different head, use the six-id comparison instead of swapping GLM “just in case”.
The model picker in Telegram does not change the API. An empty model calls GLM Flash until the client sends another id.
How to call it
Copy the id in full. Do not append a [1m] suffix. Client context window 1048576, output budget 16384 or your own max_tokens. Protocol and errors live in the docs.
GLM prices per million tokens: input $0.39, output $1.58, cache $0.16. Minimum per request $0.005. Other model cards are on pricing, not here.
About refusals
Uncensored here means fewer refusal theatrics, not “anything goes”. Why models refuse at all is in the note on abliteration. The service is for research: you own the prompt.
The Hugging Face page is origin of the weights, not a guide to this API: dealignai/GLM-5.3-Flash-UNCENSORED-FP8.
FAQ
Do I still send the GLM id?
An empty model already selects it. Send this id when you want to be explicit.
Is bot GLM the same as API GLM?
Same id. The Telegram switch still does not change what the key sends until the client sets model.
Where is the key?
Only in the bot, API button. Full path: how to start.