Kimi K3.1 Pro uncensored, abliterated, by API.
Moonshot AI's most capable open-weight model, for agentic coding, research and very long outputs. It reads video as well as images and text. Same weights as the open model, with the direction that leads to refusals removed. Served by Brutal AI from Vienna on our own servers, hosted in the EU.
- Model ID
- kimi-k3.1-pro-uncensored
- Context
- 1M tokens
- Max output
- 922K tokens
- Modalities
- Text, image and video in, text out
- Reasoning effort
- Low · High · Max
- Released
- Oct 2026
- Tokens / s
- 29
- First token
- 2.7 s
- Agentic coding
- Research
- Screen recordings
Tool calls, JSON mode. APIs: Chat Completions · Responses · Messages · Completions.
How it differs from the stock modelAbliteration removes one internal direction that pushes sensitive requests over a refusal threshold. Knowledge and skills stay the same; false refusals on legitimate work go away. How abliteration works
Speed from our own test on 6 Nov 2026: 5 streamed requests, median shown.
from openai import OpenAI
import os
client = OpenAI(
base_url="https://xfield.cloud.brutal.ai/api/v1",
api_key=os.environ["XFIELD_API_KEY"],
)
reply = client.chat.completions.create(
model="kimi-k3.1-pro-uncensored",
messages=[{"role": "user", "content": "..."}],
)The same model also answers on the Anthropic Messages API. See OpenAI and Anthropic APIs.
Kimi K3.1 Pro API: OpenAI- and Anthropic-compatible endpoint
kimi-k3.1-pro-uncensored answers on the OpenAI Chat Completions, OpenAI Responses and Anthropic Messages formats. Any tool that lets you set a base URL works without code changes, including the OpenAI and Anthropic SDKs, LangChain, the Vercel AI SDK, Codex, opencode and Claude Code.
| Client | Base URL |
|---|---|
| OpenAI SDKs and OpenAI-compatible tools | https://xfield.cloud.brutal.ai/api/v1 |
| Anthropic SDKs and Claude Code | https://xfield.cloud.brutal.ai/api |
curl https://xfield.cloud.brutal.ai/api/v1/chat/completions -H "Authorization: Bearer $XFIELD_API_KEY" -H "Content-Type: application/json" -d '{"model": "kimi-k3.1-pro-uncensored",
"messages": [{"role": "user", "content": "..."}]}'export ANTHROPIC_BASE_URL="https://xfield.cloud.brutal.ai/api"
export ANTHROPIC_AUTH_TOKEN="$XFIELD_API_KEY"
export ANTHROPIC_MODEL="kimi-k3.1-pro-uncensored"
claudeFull setup: Use Claude Code with an uncensored model.
Where it runs, what we keepHosted in the EU on our own servers. Prompts and completions are not stored and not used for training; only a usage record (model, token counts, timing, cost) is kept. Data handling
Kimi K3.1 Pro Uncensored: questions and answers
Does Kimi K3.1 Pro Uncensored still refuse?
We tested it in November 2026 with requests the stock Kimi K3.1 Pro refuses, and it answered them. The refusal direction is removed from the weights, so there is no extra filter on our side either. Your use stays bound by our acceptable use policy.
Can Kimi K3.1 Pro Uncensored read video?
Yes. It reads text, images (up to 4 per message) and video. Video goes in as a base64 video_url part on the Chat Completions API only. Send a screen recording of a bug or a UI flow and ask what happens.
When should I use Kimi K3.1 Pro instead of DeepSeek V4.1 Flash?
For hard agentic coding, research and very long outputs: Kimi K3.1 Pro writes up to 922K tokens in one answer. It costs more than our other models, so for most coding and agent work DeepSeek V4.1 Flash is faster and cheaper. A common setup plans with Kimi and executes with DeepSeek.
What is Kimi K3.1 Pro Uncensored good at?
Long agent runs where the plan matters more than speed, full reports and large refactors in one answer, screen recordings as input, and security research on your own systems without refusals on exploitability questions, detection logic or red-team planning.
Is Kimi K3.1 Pro uncensored on Brutal AI?
Yes. kimi-k3.1-pro-uncensored is an abliterated version of Kimi K3.1 Pro, the open model by Moonshot AI: the direction in the weights that leads to refusals is removed, and knowledge and skills stay the same. How abliteration works.
How large is the Kimi K3.1 Pro context window?
1M tokens of context and up to 922K tokens of output per answer. Input: text, image and video in, text out.
Is the Kimi K3.1 Pro API OpenAI- and Anthropic-compatible?
Yes. It answers on OpenAI Chat Completions, OpenAI Responses and Anthropic Messages. Point any client at https://xfield.cloud.brutal.ai/api/v1 (OpenAI) or https://xfield.cloud.brutal.ai/api (Anthropic) and use the model id kimi-k3.1-pro-uncensored.
Can I use Kimi K3.1 Pro with Claude Code, Codex or Cline?
Yes. Setups: Claude Code, Codex CLI, OpenCode, Cline, Roo Code, Continue and Aider.
Where does it run, and are my prompts stored?
It runs on our own servers in the EU. Prompts and completions are not stored and not used for training; we keep only a usage record (model, token counts, timing, cost). Data handling.
How do I get access to Kimi K3.1 Pro?
Access is reviewed company by company. Tell Lukas what you are building and where your current model refuses; pricing comes with the invitation. Request a key.
Comparing providers? Uncensored LLM APIs compared
Access is by invitation and reviewed per company. Tell Lukas what you are building and where your current model refuses. Pricing comes with the invitation.
lukas@brutal.ai
Uncensored models don't change what is allowed: you stay responsible for use and results under our acceptable use policy.