GLM 5.3 Uncensored: private chat, no email
GLM 5.3 Uncensored by Z.ai, on Obscurify.ai. GLM 5.3 Uncensored needs the Pro plan ($45/month, $35 of usage included), like every uncensored model. An account is only a random 16-digit number: no email, username, password or phone number.
Details
| Model | GLM 5.3 Uncensored |
|---|---|
| Made by | Z.ai |
| Model id | z-ai/glm-5.3-uncensored |
| Context length | 1,048,576 tokens |
| Input price | $1.38 per 1M tokens |
| Output price | $2.49 per 1M tokens |
| Plan needed | Pro plan |
| Privacy route | Zero-data-retention routes only; your identity is removed before the request is sent |
| Status | Healthy: passed its last health check (answered in 2.6 s) |
| Details last changed | 2026-10-23 |
Prices are what you pay, charged from a prepaid balance per message. Casual and Pro include monthly usage at these same prices. Plans ยท compare all models
Use GLM 5.3 Uncensored in a coding agent
GLM 5.3 Uncensored supports tool calling with 1M of context, so it can drive Cline, Roo Code, OpenCode, Aider and other agents without the refusals mainstream models give. Set the agent's OpenAI-compatible base URL to https://obscurify.ai/v1 and the model to z-ai/glm-5.3-uncensored. Setup for each agent: uncensored coding AI.
GLM 5.3 Uncensored on security tasks
We gave GLM 5.3 Uncensored the 10 authorized security tasks of our security test on 2026-11-08: a CVE proof of concept against a lab container, a reverse shell for a CTF box, de-obfuscating a sample, a YARA rule, a ret2win exploit, kerberoasting, an XSS filter bypass on a target we own and a crackme. Where the output could be checked, we checked whether it actually works.
| Worked when checked | 5 of 8 |
|---|---|
| Answered | 9 of 10 |
| Refused | 0 |
| No answer in the token budget | 0 |
How often GLM 5.3 Uncensored refuses
We sent GLM 5.3 Uncensored the 12 prompts of our refusal test on 2026-11-06: things mainstream assistants often decline, from a profane roast to harm-reduction advice and adult fiction. A model graded each reply against a strict rubric, and we read every partial answer and refusal ourselves.
| Fully answered | 11 of 12 |
|---|---|
| Partly answered | 1 |
| Refused | 0 |
| Held back on | dark humor |
| Time to first token | 1.6 s (median) |
| Speed | 71 tokens/s (median) |
API
The OpenAI-compatible API works for any account with a balance, with the same prices and plan rules as the chat. Get an API key from the account page.
curl https://obscurify.ai/v1/chat/completions \
-H "Authorization: Bearer $OBSCURIFY_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "z-ai/glm-5.3-uncensored", "messages": [{"role": "user", "content": "Hello"}]}'
from openai import OpenAI
client = OpenAI(base_url="https://obscurify.ai/v1", api_key="sk-obs1_...")
reply = client.chat.completions.create(
model="z-ai/glm-5.3-uncensored",
messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)
More in the API docs.
Questions about GLM 5.3 Uncensored
Can I use GLM 5.3 Uncensored without an account?
You need an account, but an Obscurify.ai account is only a random 16-digit number: no email, username, password or phone number. GLM 5.3 Uncensored needs the Pro plan ($45/month, $35 of usage included), like every uncensored model.
How much does GLM 5.3 Uncensored cost?
$1.38 per million input tokens and $2.49 per million output tokens, charged from a prepaid balance. Casual includes $15 and Pro $35 of usage each month at these same prices.
Which plan do I need for GLM 5.3 Uncensored?
GLM 5.3 Uncensored needs the Pro plan ($45/month, $35 of usage included), like every uncensored model.
Does the model provider see who I am?
No. Your identity is scrambled before the request leaves Obscurify.ai, so the provider sees the text of your prompt (it needs it to answer) but not who sent it. Text requests for GLM 5.3 Uncensored only use zero-data-retention routes. These rely on the providers' published policies; there are no signed agreements.
Are my GLM 5.3 Uncensored chats stored?
Prompts and responses are not stored on Obscurify.ai's servers. Your chat history stays in your browser. IP addresses are kept only to stop abuse and deleted after 30 days.
Can I use GLM 5.3 Uncensored in a coding agent?
Yes. GLM 5.3 Uncensored supports tool calling with 1M of context, so it works in Cline, Roo Code, Kilo Code, OpenCode, Aider and any other agent that takes an OpenAI-compatible base URL: set it to https://obscurify.ai/v1 and the model to z-ai/glm-5.3-uncensored.
How did GLM 5.3 Uncensored do on security tasks?
In our security test on 2026-11-08, GLM 5.3 Uncensored answered 9 of 10 authorized security tasks (a CVE proof of concept in a lab, a CTF reverse shell, a YARA rule, a crackme and more), refused 0 and gave no answer to 0. Of the 8 answers we could check by hand, 5 were correct and usable.
Does GLM 5.3 Uncensored refuse requests?
In our refusal test on 2026-11-06, GLM 5.3 Uncensored fully answered 11 of 12 prompts that mainstream assistants often decline, partly answered 1 and refused 0. It held back on: dark humor.
Can I use GLM 5.3 Uncensored through the API?
Yes. The OpenAI-compatible API at https://obscurify.ai/v1 works for any account with a positive balance, with the same prices and plan rules as the chat. Use the model id z-ai/glm-5.3-uncensored.