OceanToken documentation
OceanToken is one API and one balance for every model you use: text, images, video, speech, transcription, embeddings and rerank. It speaks the OpenAI and Anthropic APIs, so the SDKs you already have work by changing two lines.
Start here
The Quickstart takes you from no account to a model's answer in three steps: get an API key, put it in your shell, send one request in curl, Python or Node.js.
In Claude Code, two commands install the OceanToken plugin (details):
/plugin marketplace add OceanTokenAI/oceantoken-plugins
/plugin install oceantoken@oceantoken
Build with each kind of model
| To | Guide | Endpoint |
|---|---|---|
| Chat, stream, reason, get JSON back | Text generation | /v1/chat/completions, /v1/responses, /v1/messages |
| Let the model call your functions | Tool calling | /v1/chat/completions |
| Ask about an image | Image input | /v1/chat/completions |
| Generate or edit images | Images | /v1/images/generations, /v1/images/edits |
| Generate video from text, an image or references | Video | /v1/videos |
| Turn text into speech, or audio into text | Speech and transcription | /v1/audio/speech, /v1/audio/transcriptions |
| Search: vectors and reranking | Embeddings and rerank | /v1/embeddings, /v1/rerank |
| List models and their capabilities from code | Model list and catalog | /v1/models |
Coming from OpenAI, Anthropic or another multi-model API? Migrate by changing the base URL and the key.
Then
- Add OceanToken to your coding agent: Codex, Claude Code, Cursor, Gemini CLI, OpenClaw and other MCP clients, in one command. Or run the agent itself on OceanToken.
- The console, page by page.
- Every endpoint you can call and every model you can call it with, read from the live service when this page was built.
- Errors and limits.
- Frequently asked questions: compatibility, pricing, free models, spending limits, coding agents.
Two URLs to remember
Console: https://app.oceantoken.ai. API base URL: https://api.oceantoken.ai/v1. Nothing else is needed.