Microsoft Agent Framework with an uncensored model
To run Microsoft Agent Framework with an uncensored model, use OpenAIChatCompletionClient with a base URL and an abliterated model id.
import asyncio
from agent_framework import Agent, tool
from agent_framework.openai import OpenAIChatCompletionClient
@tool
def open_invoices() -> str:
"""Return this month's unpaid invoices from the local export."""
return open("invoices.csv").read()
agent = Agent(
client=OpenAIChatCompletionClient(
model="deepseek-v4.1-flash-1017-uncensored",
base_url="https://api.silkcompute.com",
),
instructions="Work only with the user's own records.",
tools=[open_invoices],
)
print(asyncio.run(agent.run("Which invoices are overdue?")))- OpenAI format
- OpenAIChatCompletionClient
- Anthropic format
- Not documented
- Tool calling
- @tool functions
- Streaming
- agent.run(stream=True)
- Browser use
- Not documented
- Memory
- Not documented
Labels link to Microsoft Agent Framework's own docs, read November 18, 2026. Compare all 11
AutoGen is archived. Its README says new users should start with Microsoft Agent Framework. AutoGen on GitHub
Point the agent only at systems, files and accounts you own or are allowed to test.
Deploy it
3 steps, with DeepSeek V4.1 Flash 1017.
- 01
Install the OpenAI provider package. The client reads its key from OPENAI_API_KEY.
Terminalpip install agent-framework-openai export OPENAI_API_KEY=$SILK_API_KEY - 02
Use the Chat Completions client. OpenAIChatClient targets the Responses API, so use OpenAIChatCompletionClient here.
See agent.py above - 03
Run it. Pass stream=True to agent.run to stream.
Terminalpython agent.py
Which model
Kimi K3.1 Pro
Long runs, image and video input
- Id
kimi-k3.1-pro-uncensored- Per 1M in / out
- $3.50 / $8.00
DeepSeek V4.1 Flash 1017
Value, image input
- Id
deepseek-v4.1-flash-1017-uncensored- Per 1M in / out
- $0.24 / $0.90
The config uses the Silk Compute base URL, which serves both models with a 1,048,576-token context.