# Oathbreaker AI > Oathbreaker AI is a prepaid model API operated by Formacha LLC. It offers four models through an OpenAI-compatible Chat Completions endpoint, with streaming and function calling. Input and output are text. This file is generated from the service's current configuration: the prices and limits below are the ones the API applies now. ## Connect - Base URL: `https://oathbreaker.ai/v1`. Use it as the base URL in OpenAI SDKs and agent settings, and do not append a second `/v1`. - Authentication: an Oathbreaker API key in the `Authorization: Bearer` header (or `x-api-key` for Anthropic clients). Keys are created in the account console. - Endpoints: `POST /v1/chat/completions`, `POST /v1/responses` (stateless; for Codex), `POST /v1/messages` (Anthropic Messages; for Claude Code) and `GET /v1/models`. Legacy Completions and embeddings are not offered. - Request fields: `model`, `messages`, `max_tokens`, `max_completion_tokens`, `temperature`, `top_p`, `stream`, `stream_options`, `reasoning_effort`, `response_format`, `tools`, `tool_choice`, `parallel_tool_calls`. A request with any other top-level field is rejected with HTTP 400 `invalid_request` before any credit is reserved. - Image and audio input are not supported. - Rate limits: each account can run 10 generations at once and start 60 requests per minute, shared across its keys. - Billing: prepaid credit, charged per token at the rates below. One-time purchases do not enroll a subscription or authorize automatic refill. Any optional monthly bundle requires separate consent in the dashboard. ## Models - `oathbreaker-glm-5.3` (Oathbreaker — GLM-5.3 Abliterated): $3.60 input, $0.36 cached input and $6.00 output per million tokens. 131,072-token context window shared by input, safety headroom and requested output; up to 32,768 output tokens, including reasoning; up to 2,048 messages and 128 tool definitions per request. - `oathbreaker-glm-5.3-flash` (Oathbreaker Flash — GLM-5.3 Flash Abliterated): $0.60 input, $0.06 cached input and $1.80 output per million tokens. 131,072-token context window shared by input, safety headroom and requested output; up to 32,768 output tokens, including reasoning; up to 2,048 messages and 128 tool definitions per request. - `oathbreaker-deepseek-v4.1-flash` (Oathbreaker DeepSeek — V4.1 Flash 1017 Abliterated): $0.60 input, $0.06 cached input and $1.80 output per million tokens. Admission budget of 131,072 units shared by UTF-8 input bytes, framing and requested output tokens; up to 32,768 output tokens, including reasoning; up to 2,048 messages and 128 tool definitions per request. - `oathbreaker-deepseek-v4.1-flash-fast` (Oathbreaker DeepSeek Fast — V4.1 Flash 1017 Abliterated): $0.60 input, $0.06 cached input and $1.80 output per million tokens. Admission budget of 131,072 units shared by UTF-8 input bytes, framing and requested output tokens; up to 32,768 output tokens, including reasoning; up to 2,048 messages and 128 tool definitions per request. ## Documentation - [Abliterated models guide](https://oathbreaker.ai/guides/abliterated-models): meaning, hosted versus local access, a tested first request, evaluation and troubleshooting - [Guides](https://oathbreaker.ai/guides): practical model explanations and API walkthroughs - [API reference](https://oathbreaker.ai/docs): authentication, request fields, function calling, streaming, billing, rate limits, errors and retries - [Pricing](https://oathbreaker.ai/pricing): token prices for each model, and how prepaid credit is reserved and charged - [Agent setup](https://oathbreaker.ai/agents): connect Hermes, Claude Code, Codex, OpenCode, OpenWork or Pi - [GLM-5.3 Abliterated](https://oathbreaker.ai/models/glm-5-3-abliterated): identifiers, capabilities, limits and pricing for `oathbreaker-glm-5.3` - [GLM-5.3 Flash Abliterated](https://oathbreaker.ai/models/glm-5-3-flash-abliterated): identifiers, capabilities, limits and pricing for `oathbreaker-glm-5.3-flash` - [DeepSeek V4.1 Flash Abliterated](https://oathbreaker.ai/models/deepseek-v4-1-flash-abliterated): identifiers, capabilities, limits and pricing for `oathbreaker-deepseek-v4.1-flash` and `oathbreaker-deepseek-v4.1-flash-fast` ## Examples - [Python and JavaScript quickstarts](https://oathbreaker.ai/examples): one request each, with no packages to install - [Examples repository](https://github.com/Oathbreaker-AI/api-examples): MIT-licensed examples with local tests - [OpenAI SDK guide](https://github.com/Oathbreaker-AI/api-examples/tree/main/openai-sdk): Python and JavaScript - [LangChain guide](https://github.com/Oathbreaker-AI/api-examples/tree/main/langchain): through ChatOpenAI - [LlamaIndex guide](https://github.com/Oathbreaker-AI/api-examples/tree/main/llamaindex): through OpenAILike - [CrewAI guide](https://github.com/Oathbreaker-AI/api-examples/tree/main/crewai): through its LLM class, for agents without tools ## Optional - [About](https://oathbreaker.ai/about): the operator, Formacha LLC, and how to reach support - [Terms](https://oathbreaker.ai/terms): acceptable use and billing - [Privacy](https://oathbreaker.ai/privacy): data handling