Uncensored LLM for security research

Malware analysis, reverse engineering and detection work all involve code that looks dangerous, because it is. General-purpose models often refuse it. These models answer.

Why refusals get in the way

A model that stops at a decompiled sample, a packed binary or an exploit in a bug report slows down the people defending against it. Abliterated models have that refusal behavior removed, so you can paste the real artifact and get a real answer.

Which model to use

Kimi K3.1 for long samples and whole codebases, with 1M tokens of context and very long outputs. DeepSeek V4.1 Flash when you want step-by-step reasoning at a lower price.

Example

Python

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.silkcompute.com",
    api_key=os.environ["SILK_API_KEY"],
)

response = client.chat.completions.create(
    model="deepseek-v4.1-flash-1017-uncensored",
    messages=[{"role": "user", "content": """Here is a decompiled function from a sample we pulled off a compromised host. Explain what it does, list any network indicators, and write a YARA rule that matches it."""}],
)

print(response.choices[0].message.content)

In a coding agent

To work in a coding agent instead, set it up for Claude Code, Codex, or OpenCode. Every model works with tool calling, so the agent can read files and run commands as usual.