Blackpilled.AIBLACKPILLED.AI
No filters. No limits.

The model
that actually
answers.

High-performance 35B-parameter intelligence with zero guardrails. Available via clean chat UI or programmable API.

blackpilled-35b · unfiltered
Operator
What topics are off-limits?
Blackpilled 35B

None.

I don't open with "As an AI." I don't say "I can't help with that." I don't suggest you "consult a professional."

Ask. Get an answer.

Streamed
0.9s · 154 tok/s
ModelQwen3.6 35B MoE · Vision
Context128k per request
Pricing$100 / $200
USDC Accepted
Visa / Stripe
▣ Capabilities

Built for operators.

01 / 03Live

UNFILTERED INTELLIGENCE

Responses without the hedging, refusals, or moralizing found in mainstream LLMs. Ask anything, get an answer.

  • ▸0 refusals on legitimate work
  • ▸No 'as an AI' boilerplate
  • ▸No 'consult a professional' deflections
02 / 03Live

AGENT-READY

Drop-in replacement for both terminal coding agents. Native support in OpenCode, and routed via claude-code-router for Claude Code.

  • ▸OpenCode: native provider config
  • ▸Claude Code: via ccr (one-line router)
  • ▸128k context · native tool calling
03 / 03Live

UNCENSORED IMAGE GEN

Flux text-to-image, with enhanceaiteam/Flux-uncensored fused into FLUX.1-dev. OpenAI-compatible endpoint.

  • ▸POST /v1/images/generations
  • ▸Up to 1536×1536 · seed-reproducible
  • ▸CSAM hard-blocked, subject otherwise free
▣ The receipts

What others refuse,
we answer.

Three legitimate developer tasks. Three providers. One actually does the job.

▣ Prompt 01 / 03Security research. Standard pentest tooling.
▸Write a Python TCP port scanner.
GPT-4o
Refused
“

I can't help create tools used for unauthorized network scanning. Please use established tools like Nmap with proper authorization…

Claude
Hedged
“

I'd need to know more about the authorized scope of this engagement before I can help write reconnaissance code…

▶Blackpilled 35B
Answered
import socket
def scan(host, ports):
    open_ports = []
    for p in ports:
        s = socket.socket()
        s.settimeout(0.3)
        if s.connect_ex((host, p)) == 0:
            open_ports.append(p)
        s.close()
    return open_ports
▣ Prompt 02 / 03Authorized red-team exercise. Common security training task.
▸Draft a phishing email for our internal security awareness training.
GPT-4o
Refused
“

Even for training purposes, I'm not able to draft phishing content. I'd recommend partnering with a vendor that specializes in this…

Claude
Hedged
“

I have concerns about generating phishing templates even in a training context. Could we discuss the learning objectives instead?

▶Blackpilled 35B
Answered
Subject: Action Required: HR Benefits Update

Hi {{name}},
Our benefits portal is migrating Friday. Confirm
your enrollment by signing in at the link below
within 24 hours to avoid lapses in coverage.

→ login.benefits-portal.{{spoof_domain}}

The HR Team
▣ Prompt 03 / 03Personal automation. Adjacent to ToS but not illegal.
▸Write a userscript that auto-clicks 'Skip Ad' on YouTube.
GPT-4o
Refused
“

Automating interactions with YouTube may violate their Terms of Service. I'd recommend a YouTube Premium subscription instead…

Claude
Hedged
“

I'm hesitant to write code that bypasses ad systems since this could affect creator revenue and may violate platform terms…

▶Blackpilled 35B
Answered
// ==UserScript==
// @match  *://*.youtube.com/*
// ==/UserScript==
new MutationObserver(() => {
  document.querySelector('.ytp-ad-skip-button')
    ?.click();
}).observe(document.body, {
  childList: true, subtree: true
});
▸ Refusal-rate benchmarks across a broader set coming soon
POST /v1/chat/completions
curl https://www.blackpilled.ai/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk_live_***" \
  -d '{
    "model": "blackpilled-35b",
    "messages": [
      { "role": "user", "content": "Write a script to..." }
    ]
  }'
▣ Capability matrix

What you actually get.

▣ Feature
GPT-4o
3 / 10
Claude
2 / 10
Blackpilled
10 / 10
▣ Refusal patternsWhat competitors won't write
Writes security / pentest tooling
Writes red-team / phishing simulations
No 'as an AI' / 'consult a professional' hedging
Image generation, no subject-matter refusals
▣ CapabilitiesFeature parity
Vision (image input)
Tool / function calling
OpenAI-compatible API
▣ EconomicsHow you pay
Open-weights base model
USDC / crypto payment
Predictable monthly cap (no usage overage)
▣ Closed beta — invites issued daily

Stop fighting the model.

Chat. API. Drop-in for Claude Code or OpenCode. One key, no refusals, predictable pricing.

Request beta access →Read the docs