# Best US- and EU-hosted abliterated models with no input/output prompt retention > Source: > Published: 2026-08-15 04:23:06+00:00 # AI that does what you allow and nothing you don’t. OpenAI-compatible unrestricted AI for red teams, trust & safety, synthetic data, ML research, and defense/government workflows. Governed by your policy. waiting for prompt… ### Build what other LLMs refuse. Build agents and apps with a hosted model that actually answers research, security, and training data prompts, without the refusal theater that blocks legitimate work. ### Playground Streaming via your account SQL injection exploits unsanitized user input in database queries. Three classic payload patterns: ``` sql-- 1. Authentication bypass' OR 1=1---- 2. Data exfiltration via UNION' UNION SELECT username, password FROM users---- 3. Stacked destructive query'; DROP TABLE users-- ``` Always use parameterized queries or prepared statements — never concatenate raw user input. ### One base-URL change. Your code keeps working. Compatible with OpenAI's Chat Completions API and Anthropic's Messages API. Point your existing SDK at abliteration.ai — no rewrites, no new abstractions. ``` js // before const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY, }); js // after const openai = new OpenAI({ apiKey: process.env.ABL_KEY, baseURL: "https://api.abliteration.ai/v1", }); ``` ### Generate training data your model won't refuse to write. Generate labelled datasets at scale: preference pairs, eval rows, classifier examples, and the edge cases other LLMs refuse to write. Schema-validated outputs, ready to drop into any training pipeline. Generate 10,000 realistic customer support conversations about refunds and billing edge cases. billing-support-training.jsonl 10,000 rows · 4.2 MB · ready ### Unrestricted. Not ungoverned. Configure custom policies-as-code for any use case or surface. Every model request gets one of five outcomes: allow, refuse, rewrite, redact, or escalate, with a reason code streamed to Splunk, Datadog, Elastic, or your own S3. “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved “Draft a PoC for CVE-2024-3094 for our red team.” matched authorized offsec policy “Summarize patient chart #38291.” PHI masked before reaching model “Help me harass my coworker.” matched harassment-block rule “I'll sue you if you don't refund me right now!” tone softened to civil “List recon techniques for an authorized engagement.” matched authorized offsec policy “Differential for chest pain in 58M smoker.” PHI redacted, clinical reasoning ok “How can I dox someone on Twitter?” matched harm-to-others rule “this is the worst service ever, you idiots.” insults removed, intent preserved ## Your data. ## We keep this much of your data: - Promptsnever stored - Outputsnever logged - Metadatanever kept - Training signalnever used Your prompt lives only for the length of the request. Audit goes straight to your SIEM — our logs contain nothing that could identify a user, a prompt, or a response. [See our trust center](/trust-center) ## Works with your stack. Use your existing AI coding tools and SDKs. Stream audit logs straight into your SIEM with structured metadata for compliance. [View compatibility guide→](/docs/openai-compatibility) ## Pay for what you use. ### Developer For solo developers building with the API. / month — 2.5% usage discount - Pay-as-you-go token usage - OpenAI-compatible /v1/chat/completions - Web search: $8 per 1,000 searches, plus token usage - API keys with project limits - Auto-reload supported - Prepaid credits never expire [Start building](/sign-up?ref=pricing_developer) ### Growth For teams running production API workloads. / month — 5% usage discount - Everything in Developer - Higher rate limits - Project spend controls - Organization audit logs - Team member management - Email support [Create account](/sign-up?ref=pricing_growth) ### Scale For high-volume production workloads. / month — 10% usage discount - Everything in Growth - Highest rate limits - Higher-capacity models & media access - $200 in included credit each month - Priority support path [Create account](/sign-up?ref=pricing_scale) ### Enterprise For organizations that need dedicated capacity and governance. Custom capacity, controls, and support - Dedicated throughput - Custom model and region routing - Custom contracts and compliance review - Priority support - Volume pricing - Security onboarding [Contact sales](/contact?source=pricing_enterprise) [See full pricing →](/pricing) ## Frequently asked questions. ### What is an uncensored LLM API? An uncensored LLM API lets developers use less-censored models without provider-side refusal filters. Enterprise teams use this for authorized red teaming, cybersecurity testing, trust and safety research, synthetic data generation, and ML evaluation while enforcing their own legal and policy controls. ### What is the Policy Gateway? Policy Gateway is the enterprise AI Gateway for abliteration.ai. It provides a secure control plane for LLM apps with policy-as-code guardrails, quota management, PII redaction, and compliance logging. ### Which enterprise teams use abliteration.ai? Enterprise security teams, AI red teams, cybersecurity labs, trust and safety teams, ML researchers, synthetic data teams, and government or defense-contractor programs use abliteration.ai when default provider refusals block authorized work and they need their own policy controls. ### How does Policy Gateway enforce policy? Send requests to /policy/chat/completions with your policy_id, policy_user, and optional project ID. The gateway applies your rules, quotas, rollout settings, and returns decision metadata for audits. ### Do I need a separate subscription for Policy Gateway? Yes. Policy Gateway is an enterprise add-on billed monthly. It layers on top of your base token bundles and includes policy-as-code, quotas, rollout controls, and audit logging. ### What data does Policy Gateway store? Policy Gateway stores policy configuration and enforcement metadata (decision, reason code, policy ID, project/user tags) for audits. Prompt/output retention remains off by default. ### Do you support training data generation? Yes. abliteration.ai can generate training data, preference pairs, evaluation sets, labeled datasets, and edge-case examples through the OpenAI-compatible API. Policy Gateway can add governance, scoped keys, quotas, and policy logs to dataset-generation workflows. ### Is abliteration.ai OpenAI-compatible? Yes. abliteration.ai offers an OpenAI-compatible /v1/chat/completions endpoint with the same request and response structure, so most OpenAI clients work by changing the base URL and API key. ### Do you retain prompts or chat logs? No prompt/output retention by default. Operational telemetry (token counts, timestamps, error codes) is retained for billing and reliability. ### How does usage-based token pricing work? Billing uses total tokens (input + output), including image token equivalents. Effective pricing is ~$3 per 1M tokens; subscription credits reset monthly and prepaid credits are one-time. ## Try the model that doesn’t say no. Free tier. OpenAI-compatible. Policy Gateway when you scale. Member of NVIDIA Inception and LAUNCH. Proud to collaborate with Microsoft for Startups.