# abliteration.ai

> abliteration.ai is an inference API for unrestricted open-weight models, compatible with the OpenAI and Anthropic SDKs, with a built-in policy gateway.

## Docs

- [Count input tokens](https://docs.abliteration.ai/api-reference/anthropic-messages/count-input-tokens.md): Returns an estimated input token count for a message payload without generating a response. Useful for budget checks before sending expensive prompts.
- [Create a message](https://docs.abliteration.ai/api-reference/anthropic-messages/create-a-message.md): Creates a message using the Anthropic Messages API format. Set stream: true for server-sent events.
- [Create a chat completion](https://docs.abliteration.ai/api-reference/chat-completions/create-a-chat-completion.md): Creates a chat completion. Set stream: true for server-sent events.
- [List models](https://docs.abliteration.ai/api-reference/models/list-models.md): Returns the list of available models.
- [Create a response](https://docs.abliteration.ai/api-reference/responses/create-a-response.md): Creates a response using the OpenAI Responses API format. Set stream: true for server-sent events.
- [Use the Anthropic SDK with abliteration.ai](https://docs.abliteration.ai/api/anthropic-compatibility.md): How to use abliteration.ai with the Anthropic Messages API — base URL, auth token, and what's supported.
- [abliteration.ai API errors](https://docs.abliteration.ai/api/errors.md): HTTP error codes, error body shapes (OpenAI and Anthropic), and how to handle policy-blocked requests.
- [Introduction](https://docs.abliteration.ai/api/introduction.md): Overview of abliteration.ai's HTTP APIs — base URL, authentication, and how the OpenAI-compatible and Anthropic-compatible surfaces are exposed.
- [Use any OpenAI SDK with abliteration.ai](https://docs.abliteration.ai/api/openai-compatibility.md): How to use abliteration.ai with any OpenAI SDK — base URL, authentication, supported features, and caveats.
- [Policy endpoints](https://docs.abliteration.ai/api/policy-endpoints.md): The /policy/* API surface that adds project quotas, policy evaluation, streaming metadata, and audit events.
- [How to authenticate with abliteration.ai](https://docs.abliteration.ai/authentication.md): How to authenticate abliteration.ai API requests with bearer tokens, scope keys to projects, and rotate keys safely.
- [Count input tokens before sending](https://docs.abliteration.ai/capabilities/count-tokens.md): Measure the exact input-token cost of a request before sending it, via POST /v1/messages/count_tokens.
- [Send images to abliteration.ai](https://docs.abliteration.ai/capabilities/images.md): Send images alongside text on the OpenAI Chat Completions and Anthropic Messages surfaces.
- [Stream tokens from abliteration.ai](https://docs.abliteration.ai/capabilities/streaming.md): Stream abliteration.ai responses as server-sent events to reduce time-to-first-token.
- [Streaming policy metadata](https://docs.abliteration.ai/capabilities/streaming-policy-metadata.md): On `/policy/*` endpoints, abliteration.ai injects a policy object into every streaming frame so clients can render compliance UI.
- [Toggle thinking on abliteration.ai](https://docs.abliteration.ai/capabilities/thinking.md): Toggle the thinking step on abliteration.ai with the thinking parameter on Chat Completions and Anthropic Messages.
- [Tool calling on abliteration.ai](https://docs.abliteration.ai/capabilities/tool-calling.md): Function calling on abliteration.ai across all three API surfaces. Pick the format that matches your client.
- [Anthropic Messages](https://docs.abliteration.ai/capabilities/tool-calling/anthropic-messages.md): Tool calling on /v1/messages — input_schema definition, tool_use response blocks, tool_result continuation blocks.
- [OpenAI Chat Completions](https://docs.abliteration.ai/capabilities/tool-calling/openai-chat-completions.md): Tool calling on /v1/chat/completions — define tools, handle the call/execute/continue loop, and stream tool arguments.
- [OpenAI Responses API](https://docs.abliteration.ai/capabilities/tool-calling/openai-responses.md): Tool calling on /v1/responses — flat function schema, function_call items in output[], function_call_output continuation.
- [Send video to abliteration.ai](https://docs.abliteration.ai/capabilities/video.md): Send short video clips alongside text on the OpenAI Chat Completions surface, either base64-inlined or as an HTTPS URL.
- [Web fetch on abliteration.ai](https://docs.abliteration.ai/capabilities/web-fetch.md): Server-side web fetch on abliteration.ai — let the model pull and read a specific URL via the Anthropic Messages API.
- [Web search on abliteration.ai](https://docs.abliteration.ai/capabilities/web-search.md): Enable web search on abliteration.ai across OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages — citations included.
- [Anthropic Messages](https://docs.abliteration.ai/capabilities/web-search/anthropic-messages.md): Server-side web search on /v1/messages via the tools array, with native allowed_domains and blocked_domains at the tool's top level.
- [OpenAI Chat Completions](https://docs.abliteration.ai/capabilities/web-search/openai-chat-completions.md): Server-side web search on /v1/chat/completions via the web_search_options request field.
- [OpenAI Responses API](https://docs.abliteration.ai/capabilities/web-search/openai-responses.md): Server-side web search on /v1/responses via the tools array, with OpenAI's native tools_options.web_search.filters.include_domains allow list.
- [Compatibility matrix](https://docs.abliteration.ai/compatibility-matrix.md): What abliteration.ai supports across OpenAI Chat Completions, Responses, Anthropic Messages, and Count Tokens surfaces.
- [FAQ](https://docs.abliteration.ai/faq.md): Frequently asked questions about abliteration.ai — compatibility, models, retention, and the Policy Gateway.
- [abliteration.ai documentation](https://docs.abliteration.ai/index.md): abliteration.ai is an inference API for unrestricted, uncensored open-weight models, compatible with the OpenAI and Anthropic SDKs, with a built-in policy gateway.
- [Use Claude Code with abliteration.ai](https://docs.abliteration.ai/integrations/claude-code.md): Route Anthropic's Claude Code CLI through abliteration.ai by setting two auth variables and picking how to surface abliterated-model.
- [Use Claude Cowork with abliteration.ai](https://docs.abliteration.ai/integrations/claude-cowork.md): Configure Claude Cowork on 3P (Claude Desktop's third-party inference mode) to route through abliteration.ai.
- [Cloudflare Workers](https://docs.abliteration.ai/integrations/cloudflare-workers.md): Call abliteration.ai from any Cloudflare Worker with fetch — no SDK required.
- [Use OpenAI Codex with abliteration.ai](https://docs.abliteration.ai/integrations/codex.md): Use OpenAI's Codex CLI with abliteration.ai by registering a custom provider in ~/.codex/config.toml.
- [LangChain](https://docs.abliteration.ai/integrations/langchain.md): Drop-in abliteration.ai with LangChain and LangGraph via the standard ChatOpenAI class.
- [LlamaIndex](https://docs.abliteration.ai/integrations/llamaindex.md): Use abliteration.ai as the LLM in LlamaIndex pipelines via the OpenAILike adapter.
- [Node / TypeScript](https://docs.abliteration.ai/integrations/node.md): Use the official OpenAI or Anthropic JavaScript SDKs with abliteration.ai in Node, Edge, and Bun runtimes.
- [Use OpenClaw with abliteration.ai](https://docs.abliteration.ai/integrations/openclaw.md): Use OpenClaw (open-source agent framework, 100+ skills) with abliteration.ai.
- [Python](https://docs.abliteration.ai/integrations/python.md): Use the official OpenAI or Anthropic Python SDKs with abliteration.ai by changing the base URL.
- [Vercel AI SDK](https://docs.abliteration.ai/integrations/vercel-ai-sdk.md): Wire abliteration.ai into the Vercel AI SDK using the OpenAI-compatible provider.
- [abliteration.ai models](https://docs.abliteration.ai/models.md): Available abliteration.ai models — `abliterated-model` is an unrestricted, uncensored open-weight LLM with vision and tool use.
- [Connectors](https://docs.abliteration.ai/policy-gateway/connectors.md): Stream Policy Gateway events to your SIEM, log pipeline, or data lake. Thirteen destinations available.
- [Integration](https://docs.abliteration.ai/policy-gateway/integration.md): Wire Policy Gateway into your request flow and read policy decisions on the client.
- [Onboarding](https://docs.abliteration.ai/policy-gateway/onboarding.md): Set up your first policy, project, and scoped key on Policy Gateway in six steps.
- [Policy Gateway overview](https://docs.abliteration.ai/policy-gateway/overview.md): Policy Gateway is abliteration.ai's governance layer — rules, rollout modes, and audit events for every request.
- [Azure Blob Storage](https://docs.abliteration.ai/policy-gateway/policy-logs/azure-blob.md): Archive Policy Gateway events in an Azure Blob Storage container.
- [Azure Monitor](https://docs.abliteration.ai/policy-gateway/policy-logs/azure-monitor.md): Send Policy Gateway events to an Azure Log Analytics workspace via the Data Collector API.
- [Backblaze B2](https://docs.abliteration.ai/policy-gateway/policy-logs/backblaze-b2.md): Archive Policy Gateway events in a Backblaze B2 bucket via the S3-compatible API.
- [Cloudflare R2](https://docs.abliteration.ai/policy-gateway/policy-logs/cloudflare-r2.md): Archive Policy Gateway events in Cloudflare R2 with zero egress fees.
- [Datadog Logs](https://docs.abliteration.ai/policy-gateway/policy-logs/datadog.md): Stream Policy Gateway events to Datadog Logs via the intake API.
- [Elastic / OpenSearch](https://docs.abliteration.ai/policy-gateway/policy-logs/elastic.md): Index Policy Gateway events into Elasticsearch or Elastic Cloud via the Bulk API.
- [Google Cloud Storage](https://docs.abliteration.ai/policy-gateway/policy-logs/gcs.md): Archive Policy Gateway events in Google Cloud Storage via S3-compatible HMAC keys.
- [HTTP Webhook](https://docs.abliteration.ai/policy-gateway/policy-logs/http.md): POST Policy Gateway events as NDJSON to any HTTPS endpoint — Slack, PagerDuty, internal services, Zapier.
- [OpenTelemetry](https://docs.abliteration.ai/policy-gateway/policy-logs/otel.md): Export Policy Gateway events as OTLP log records over HTTP or gRPC to any OpenTelemetry collector.
- [Amazon S3](https://docs.abliteration.ai/policy-gateway/policy-logs/s3.md): Archive Policy Gateway events in an Amazon S3 bucket as JSON objects for long-term retention.
- [S3-Compatible](https://docs.abliteration.ai/policy-gateway/policy-logs/s3-compatible.md): Archive Policy Gateway events in any S3-protocol storage — MinIO, Wasabi, DigitalOcean Spaces.
- [Splunk HEC](https://docs.abliteration.ai/policy-gateway/policy-logs/splunk-hec.md): Stream Policy Gateway events to Splunk via HTTP Event Collector.
- [Security](https://docs.abliteration.ai/policy-gateway/security.md): How Policy Gateway handles your data — retention, encryption, access control.
- [abliteration.ai quickstart](https://docs.abliteration.ai/quickstart.md): Get your first abliteration.ai request working in under a minute — grab a key, hit /v1/chat/completions, stream tokens.
- [What is abliteration?](https://docs.abliteration.ai/what-is-abliteration.md): Abliteration is a weight-modification technique that removes the refusal direction from an open-weight LLM, producing an unrestricted model that responds to prompts the original would refuse.

## OpenAPI Specs

- [openapi](https://docs.abliteration.ai/api-reference/openapi.json)