Kimi K3.1 Pro for AI agents

Kimi K3.1 Pro is the model to put behind an unrestricted AI agent when the run is long. It reads 1,048,576 tokens, writes up to 943,718 in one reply and takes text, images and video.

Updated November 18, 2026By unrestrictedaiagent
First tool call
curl https://api.silkcompute.com/chat/completions \
  -H "Authorization: Bearer $SILK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k3.1-pro-uncensored",
    "messages": [{"role": "user", "content": "Is the disk on my server almost full?"}],
    "tools": [{"type": "function", "function": {
      "name": "run_command",
      "description": "Run a shell command on my own server",
      "parameters": {"type": "object",
        "properties": {"cmd": {"type": "string"}}, "required": ["cmd"]}}}]
  }'

Specs

Maker
Moonshot AI
Parameters
2,780B
Context
1,048,576 tokens
Max output
943,718 tokens
Input
Text, images, video
Per 1M tokens
$3.50 input, $0.35 cached, $8.00 output
Reasoning effort
Low, high or max
Model id
kimi-k3.1-pro-uncensored

Refusal behavior is removed, and models are served without content filters.

Cost of a run

Each step is 20,000 input and 1,000 output tokens, at prices read November 18, 2026.

50 steps
No cache
$3.90
80% cached
$1.38
200 steps
No cache
$15.60
80% cached
$5.52
1,000 steps
No cache
$78.00
80% cached
$27.60

Frameworks

It runs in all 11 frameworks on this site. OpenAI Agents SDK, LangGraph, CrewAI, Pydantic AI, browser-use, Hermes Agent, Goose, n8n AI Agent, Letta use it in their setup.

All 11 setups