Cookbook

Every recipe runs verbatim against https://uncensoredai.dev/v1 — replace the key and go.

Streaming responses

Token-for-token SSE deltas in the OpenAI format — set stream:true and iterate.

Streaming responses

from openai import OpenAI

client = OpenAI(base_url="https://uncensoredai.dev/v1", api_key="YOUR_KEY")

stream = client.chat.completions.create(
    model="deepseek-v4.1-flash",
    messages=[{"role": "user", "content": "Write a haiku about databases."}],
    stream=True,
)
for chunk in stream:
    delta = chunk.choices[0].delta.content
    if delta:
        print(delta, end="", flush=True)

Function calling

The model returns tool_calls JSON; you execute and send the tool result back in the same wire format.

Function calling

from openai import OpenAI
import json

client = OpenAI(base_url="https://uncensoredai.dev/v1", api_key="YOUR_KEY")

tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "Current weather for a city",
        "parameters": {
            "type": "object",
            "properties": {"city": {"type": "string"}},
            "required": ["city"],
        },
    },
}]

messages = [{"role": "user", "content": "Weather in Hong Kong?"}]
res = client.chat.completions.create(model="deepseek-v4.1-flash", messages=messages, tools=tools)

call = res.choices[0].message.tool_calls[0]
args = json.loads(call.function.arguments)
weather = {"city": args["city"], "temp_c": 28, "sky": "humid"}  # your real lookup here

messages.append(res.choices[0].message)
messages.append({"role": "tool", "tool_call_id": call.id, "content": json.dumps(weather)})

final = client.chat.completions.create(model="deepseek-v4.1-flash", messages=messages)
print(final.choices[0].message.content)

Structured JSON output

response_format json_object asks the model for machine-parseable output.

Structured JSON output

from openai import OpenAI

client = OpenAI(base_url="https://uncensoredai.dev/v1", api_key="YOUR_KEY")

res = client.chat.completions.create(
    model="deepseek-v4.1-flash",
    messages=[{
        "role": "user",
        "content": "Extract {name, severity, component} from: 'Checkout 500s when payment service times out.'",
    }],
    response_format={"type": "json_object"},
)
print(res.choices[0].message.content)

RAG grounding

Bring your own retrieval; the pattern is retrieve → cite → answer. Works with any vector store.

RAG grounding

from openai import OpenAI

client = OpenAI(base_url="https://uncensoredai.dev/v1", api_key="YOUR_KEY")

# 1. retrieve (your vector store / BM25 / SQL — anything that ranks)
chunks = search_index("our refund window")[:5]

# 2. ground the answer
context = "\n\n".join(f"[{i+1}] {c}" for i, c in enumerate(chunks))
res = client.chat.completions.create(
    model="deepseek-v4.1-flash",
    messages=[
        {"role": "system", "content": "Answer only from the context. Cite as [n]."},
        {"role": "user", "content": f"Context:\n{context}\n\nQuestion: what is our refund window?"},
    ],
)
print(res.choices[0].message.content)

Vision (image input)

Gemini accepts image_url content parts with base64 or public URLs.

Vision (image input)

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://uncensoredai.dev/v1", apiKey: "YOUR_KEY" });

const res = await client.chat.completions.create({
  model: "gemini-4-argon",
  messages: [{
    role: "user",
    content: [
      { type: "text", text: "What is in this image?" },
      { type: "image_url", image_url: { url: "https://example.com/diagram.png" } },
    ],
  }],
});
console.log(res.choices[0].message.content);