Standard LLMs decide what's allowed - refusing fiction with conflict, security research, medical questions, and more. Abliteration.ai puts you in charge with an OpenAI-compatible API, built-in chat, and a Policy Gateway to define your own rules. Set your own limits and stop getting blocked on legitimate work.
Drop-in replacement for OpenAI API. Change the base URL and keep your existing code. Works with all major SDKs.
Access less-restricted models that don't refuse legitimate requests. Perfect for creative writing, research, and specialized applications.
Define your own safety rules with policy-as-code. Enforce rewrite, redact, or escalate outcomes instead of blanket refusals.
Export audit logs to Splunk, Datadog, Elastic, and more. Track every decision with structured metadata for compliance.
~$5 per 1M tokens with no hidden fees. Prepaid credits never expire. Simple, predictable billing for your AI workloads.
Migrate from OpenAI, Azure, Anthropic, or any provider in minutes. Our migration tool patches your code automatically.
Live Console
Stream responses, attach images, and inspect the request body.
Developer-controlled, less-censored model delivered via API. We do not apply provider-side refusal filters; you control outputs and policy enforcement. Uncensored model available for teams that need them. You are responsible for keeping usage lawful in your jurisdiction—do not generate or distribute illegal content.
abliterated-modelView model specsApply one policy across apps, models, and agents with policy-as-code rules, rollout controls, per-project keys, quotas, and audit logs. Export audits to Splunk, Datadog, Elastic, S3, and Azure Monitor. Built for enterprise AI governance.
Full /v1/chat/completions parity. Works with most clients that speak this format—just point them at this base URL.
Send images alongside text prompts to extract information, summarize what’s on screen, and answer questions about photos, charts, and UI screenshots. Use the same /v1/chat/completions interface with multimodal message parts for a single, developer-friendly image understanding API.
Effective price: ~$5 per 1M tokens. Billing uses total tokens (input + output), and image inputs count as input tokens. Subscription credits reset monthly; prepaid credits do not expire. Credits are just the billing unit— see API pricing.
Default policy: no prompt/output retention. Payloads are processed transiently and never used for training. Operational telemetry (token counts, timestamps, error codes) is retained for billing and reliability.
Requests are rate-limited per API key. If you exceed limits, you will receive a 429 response with a Retry-After header. Use backoff and retries, and upgrade for priority throughput when you need more capacity.
Create an account to save credits and API keys. No phone verification required.
Generate keys for programmatic access.
Support
Quick answers about billing, integrations, and Policy Gateway governance.
An uncensored LLM API gives developers access to less-censored models without provider-side refusal filters. Uncensored models are available when needed, and you control prompts, outputs, and policy enforcement; content must still comply with your local laws and policies.
Policy Gateway is an enterprise AI governance layer for abliteration.ai. It applies policy-as-code rules, quotas, rollout controls, and audit logs across apps, models, and agents.
Send requests to /policy/chat/completions with your policy_id, policy_user, and optional project ID. The gateway enforces your rules and returns decision metadata for audits.
Yes. Policy Gateway is an enterprise add-on billed monthly. It layers on top of your base token bundles.
It stores policy configuration and enforcement metadata (decision, reason code, policy ID, project/user tags) for audits. Prompt/output retention remains off by default.
Yes. We expose an OpenAI-compatible /v1/chat/completions endpoint with the same request/response shape, so most OpenAI clients work by changing the base URL and key.
No. We do not retain prompts or outputs by default. Operational telemetry (token counts, timestamps, error codes) is retained for billing and reliability.
You’re billed on total tokens processed (input + output). Image inputs are metered as token equivalents and count as input tokens. Effective pricing is ~$5 per 1M tokens; use monthly subscription credits or one-time prepaid credits (subscriptions reset monthly with no rollover).