Uncensored Qwen3.8-27B.
Pay per token. Nothing logged.
An OpenAI-compatible API for Qwen3.8-27B with refusals removed (abliteration). 128K tokens of context, tools, reasoning with adjustable effort, and real prompt caching billed at the cached-input rate.
Create an account Read the docs
Exact model
Deployment q38u-nvfp4-64f45f2-vllm0.32.0-rtxpro6000-r1, revision 64f45f22a357, NVFP4.
Prices
$0.6 input, $0.1 cached input, $2.25 output per 1M tokens.
No prompt archive
Prompts and completions are not logged or stored. Billing keeps token counts only.
Fair billing
Engine faults are never billed. Cancelled streams pay only for tokens actually generated.
Drop-in for OpenAI SDKs
from openai import OpenAI
client = OpenAI(base_url="https://api.infrenis.com/v1", api_key="sk-inf-...")
r = client.chat.completions.create(
model="qwen3.8-27b-uncensored",
messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)