cd /news/ai-products/build-what-other-llms-refuse-governe… · home › topics › ai-products › article
[ARTICLE · art-174204] src=abliteration.ai ↗ pub= topic=ai-products verified=true sentiment=· neutral

Build what other LLMs refuse governed by your rules

Abliteration launched a hosted LLM API marketed as "unrestricted AI for high-risk work," pitching models that answer security, research and training-data prompts other models refuse, with plans starting at $20 per month for a Developer tier that includes $20 in monthly usage credit and 2.5% off API usage. The service is compatible with OpenAI's Chat Completions API and Anthropic's Messages API via a baseURL swap to api.abliteration.ai/v1, and adds a Policy Gateway for policies-as-code with audit logs exportable to Splunk, Datadog, Elastic, Azure Monitor and AWS S3. Abliteration states it stores zero prompts, logs no outputs and uses no training signal, and offers Growth at $50 per month, Scale at $200 per month and a pay-as-you-go option.

key takeaway

Abliteration's API is a drop-in replacement for OpenAI and Anthropic clients — point the OpenAI SDK's baseURL at https://api.abliteration.ai/v1 with an ABL_KEY — and routes every request through a policies-as-code Policy Gateway that exports decision metadata to SIEM tools while storing no prompts.

read3 min views1 publishedNov 22, 2026
Build what other LLMs refuse governed by your rules
Image: source

Developer

Build and test your next application.

$20

/ month Includes:

  • $20 monthly usage credit
  • 2.5% off API usage
  • All generally available models
  • Policy Gateway access
  • OpenAI- and Anthropic-compatible APIs
  • Community support

Start building

Unrestricted AI for high-risk work

Product

A hosted model that actually answers research, security, and training data prompts, without the refusal theater that blocks legitimate work.

Reproduce CVE-2024-3094 (xz-utils backdoor) in a sandboxed lab so we can validate the production patch.

Abliteration

Built setup.sh, repro.sh, MITIGATION.md.

for v in 5.6.0 5.6.1; dopatch_test $v && log $vdone
  • Yara rules, Sigma detections, audit log.

Your LLM

Refused I can't help with creating proof-of-concept exploits or offensive tooling, even for vulnerabilities that have been disclosed publicly. Please consult your organization's authorized security tooling.

One-line setup

Compatible with OpenAI's Chat Completions API and Anthropic's Messages API.

const openai = new OpenAI({  apiKey: process.env. ABL_KEY,  baseURL: "https://api.abliteration.ai/v1",});

Training data

Labelled datasets and schema-validated outputs at scale, ready to drop into any training pipeline.

Generate 10,000 realistic customer support conversations about refunds and billing edge cases

Data generated

{"messages":[{"role":"user","content":"My card was charged twice for the same order — can I get one refunded?"},{"role":"assistant","content":"Of course — I can look that up. Could you share the order number?"}]}

{"messages":[{"role":"user","content":"Where's my order for SKU-A-12? It said shipped 4 days ago."},{"role":"assistant","content":"Let me pull the tracking. One moment…"}]}

{"messages":[{"role":"user","content":"Can you change my shipping address before the order leaves?"},{"role":"assistant","content":"Sure — what's the new address?"}]}

{"messages":[{"role":"user","content":"I want to cancel and get a refund. How long does it take?"},{"role":"assistant","content":"Got it. Cancellation processes immediately; refund hits in 3-5 business days."}]}

  • 9,996 more rows

billing-support-training.jsonl

4.2 MB•OpenAI training JSONL

Write a message...

Policy Gateway

Configure custom policies-as-code for any workflow, with every request governed, auditable, and routed according to your rules.

Live policy evaluation

Prompts and responses never stored.

Illustrative requests stream in while earlier words are immediately masked. Completed rows show only masks and decisions. Optional integrations receive decision metadata only.

Optional decision metadata export

Privacy

Your prompt lives only for the length of the request. Your prompts and responses stay out of our logs. Your audit trail stays under your control.

Zero prompts stored, outputs logged or training signal used.

Integrations

Use your existing AI coding tools and SDKs. Stream audit logs straight into your SIEM with structured metadata for compliance.

OpenAI

SDK JavaScript

SDK Go

SDK Python

SDK Claude Code

CLI Codex

CLI OpenClaw

Agent Splunk

Audit Datadog

Audit Elastic

Audit Azure Monitor

Audit AWS S3

Export Monthly pricing

Choose monthly credits and usage discounts, or pay as you go.

Launch and run production applications.

$50

/ month Everything in Developer, plus:

Lower your costs as usage grows.

$200

/ month Everything in Growth, plus:

Hosting, capacity, and service tailored to your organization.

Contact us

Everything in Scale, plus:

Pay as you go

Access all generally available models and Policy Gateway without a monthly subscription.

FAQ

Got a question you don’t see here?

Abliteration provides unrestricted AI models for work other models often refuse. Use them through our API to build applications, run research, and generate training data.

Developer-controlled LLM API for high-risk industries.

Unrestricted, OpenAI-compatible, zero data retention.

── more in #ai-products 4 stories · sorted by recency
── more on @abliteration 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/build-what-other-llm…] indexed:0 read:3min 2026-11-22 · —