# OceanToken > OceanToken is one API and one prepaid balance for 400+ AI models: text (chat/reasoning), image, video, speech, transcription, embeddings and rerank, from 67 publishers including OpenAI, Anthropic, Google, DeepSeek, Qwen and ByteDance. The API is OpenAI-compatible (base URL https://api.oceantoken.ai/v1); Claude models also answer on the Anthropic Messages API (the Anthropic SDK takes https://api.oceantoken.ai and adds /v1 itself). Prices are per model, in US dollars, with no platform fee. Console: https://app.oceantoken.ai. Product site: https://oceantoken.ai. MCP server for Codex, Claude Code, Cursor and Gemini CLI: https://mcp.oceantoken.ai/mcp (install guide: https://mcp.oceantoken.ai/install). ## Documentation - [OceanToken documentation](https://docs.oceantoken.ai/): OceanToken is one OpenAI-compatible API and one prepaid balance for nearly 500 AI models — text, images, video, speech, transcription, embeddings and rerank. Create an account, add credits, create a key, make your first request. - [Add credits](https://docs.oceantoken.ai/get-started/add-credits): Add credits to your prepaid OceanToken balance by card, see where the balance shows, and how an organization shares one wallet. - [Create an account](https://docs.oceantoken.ai/get-started/create-account): Sign up for OceanToken as a person or an organization, activate the account by email, and sign in. - [Create an API key](https://docs.oceantoken.ai/get-started/create-api-key): Create an OceanToken API key with its own credit limit and expiry, send it as a bearer token, and rotate or revoke it. - [Your first request](https://docs.oceantoken.ai/get-started/first-request): Call OceanToken from curl, Python or TypeScript with the base URL https://api.oceantoken.ai/v1, stream answers, and use Claude through the Anthropic SDK. - [Profile](https://docs.oceantoken.ai/console/account): Change your email, download a copy of your data, close your OceanToken account, reset your password and sign out. - [Activity](https://docs.oceantoken.ai/console/activity): OceanToken spend, requests and tokens by day, by model and by key, over the last 7 or 30 days. - [API](https://docs.oceantoken.ai/console/api): The console's API page: the base URL, a runnable quickstart for any model, and the endpoints your account can call. - [API Keys](https://docs.oceantoken.ai/console/api-keys): Create, edit, rotate and delete OceanToken API keys in the console, each with its own credit limit, reset period and expiry. - [Chat](https://docs.oceantoken.ai/console/chat): Chat in the OceanToken console: talk to any text model, create images and videos, run deep research and work with files, on your API balance. - [Credits](https://docs.oceantoken.ai/console/credits): Your OceanToken balance, holds for jobs still running, adding credits, and organization wallets with per-team allocations. - [Logs](https://docs.oceantoken.ai/console/logs): Every request made with your OceanToken keys: time, model, tokens, cost, status, latency and the request and response bodies. - [Models](https://docs.oceantoken.ai/console/models): Browse every OceanToken model in the console, filter by type, and read each price in the unit it is billed in. - [Overview](https://docs.oceantoken.ai/console/overview): The OceanToken console's Overview page: this week's spend, requests and tokens, and quick actions for keys, models, credits and chat. - [Add OceanToken to your coding agent](https://docs.oceantoken.ai/agents/connect): One command gives Codex, Claude Code, Cursor, Gemini CLI, OpenClaw and other MCP clients OceanToken's tools for images, video, speech, transcription and other models. - [Run your coding agent on OceanToken](https://docs.oceantoken.ai/agents/run-on-oceantoken): Point Codex or Claude Code at OceanToken so the agent's own model runs on your OceanToken key and balance. - [API reference](https://docs.oceantoken.ai/api): OceanToken API reference: base URL, the OpenAI, Anthropic and Cohere request shapes, every endpoint, streaming, video jobs, idempotency and request cost. - [Errors](https://docs.oceantoken.ai/errors): OceanToken error codes and what to do about each, the request ID, when to retry, idempotency keys for paid requests, and errors while streaming. - [Frequently asked questions](https://docs.oceantoken.ai/faq): Straight answers about OceanToken — what it is, OpenAI compatibility, which of the nearly 500 models you can call, pricing per token, image, second or minute, free models, spending limits per API key, and how to connect Claude Code, Codex, Cursor or Gemini CLI. - [Limits](https://docs.oceantoken.ai/limits): OceanToken spending limits, per-key rate limits and their headers, request size, the 240-second timeout, data retention and model access. - [Models and prices](https://docs.oceantoken.ai/models): Every AI model you can call through the OceanToken API — nearly 500 text, image, video, speech, transcription, embedding and rerank models — with the exact model ID, price per 1M tokens (input, cached input, cache write, output), image, megapixel, second or minute, context window and capabilities. Read from the live service at build time. - [Retries and automatic failover](https://docs.oceantoken.ai/reliability): What OceanToken does when a model's channel fails or is busy, what it does not retry, how streaming and billing behave, and how to name your own backup models. ## Models - [All models with prices](https://docs.oceantoken.ai/models): 483 models; each has its own page at https://docs.oceantoken.ai/models/ with price, context window, capabilities and schema.org data - [Aion Labs models](https://docs.oceantoken.ai/models/aion-labs): 4 models - [Alibaba models](https://docs.oceantoken.ai/models/alibaba): 6 models - [Amazon models](https://docs.oceantoken.ai/models/amazon): 4 models - [Anthracite models](https://docs.oceantoken.ai/models/anthracite-org): 1 models - [Anthropic models](https://docs.oceantoken.ai/models/anthropic): 19 models - [Arcee AI models](https://docs.oceantoken.ai/models/arcee-ai): 1 models - [BAAI models](https://docs.oceantoken.ai/models/baai): 3 models - [Black Forest Labs models](https://docs.oceantoken.ai/models/black-forest-labs): 7 models - [ByteDance models](https://docs.oceantoken.ai/models/bytedance): 6 models - [ByteDance Seed models](https://docs.oceantoken.ai/models/bytedance-seed): 9 models - [Canopylabs models](https://docs.oceantoken.ai/models/canopylabs): 1 models - [Cognitivecomputations models](https://docs.oceantoken.ai/models/cognitivecomputations): 1 models - [Cohere models](https://docs.oceantoken.ai/models/cohere): 8 models - [Deepgram models](https://docs.oceantoken.ai/models/deepgram): 2 models - [DeepSeek models](https://docs.oceantoken.ai/models/deepseek): 17 models - [Dots Studio models](https://docs.oceantoken.ai/models/dots-studio): 1 models - [Fish Audio models](https://docs.oceantoken.ai/models/fish-audio): 5 models - [Google models](https://docs.oceantoken.ai/models/google): 38 models - [Gryphe models](https://docs.oceantoken.ai/models/gryphe): 1 models - [Hexgrad models](https://docs.oceantoken.ai/models/hexgrad): 1 models - [Heygen models](https://docs.oceantoken.ai/models/heygen): 1 models - [IBM Granite models](https://docs.oceantoken.ai/models/ibm-granite): 2 models - [Inception models](https://docs.oceantoken.ai/models/inception): 2 models - [inclusionAI models](https://docs.oceantoken.ai/models/inclusionai): 2 models - [Inference.net models](https://docs.oceantoken.ai/models/inference-net): 2 models - [intfloat models](https://docs.oceantoken.ai/models/intfloat): 3 models - [Krea models](https://docs.oceantoken.ai/models/krea): 3 models - [Kuaishou models](https://docs.oceantoken.ai/models/kwaivgi): 3 models - [Liquid AI models](https://docs.oceantoken.ai/models/liquid): 2 models - [Mancer models](https://docs.oceantoken.ai/models/mancer): 1 models - [Meituan models](https://docs.oceantoken.ai/models/meituan): 1 models - [Meta models](https://docs.oceantoken.ai/models/meta): 5 models - [Meta models](https://docs.oceantoken.ai/models/meta-llama): 8 models - [Microsoft models](https://docs.oceantoken.ai/models/microsoft): 10 models - [MiniMax models](https://docs.oceantoken.ai/models/minimax): 13 models - [Mistral AI models](https://docs.oceantoken.ai/models/mistralai): 24 models - [Moonshot AI models](https://docs.oceantoken.ai/models/moonshotai): 8 models - [Morph models](https://docs.oceantoken.ai/models/morph): 2 models - [Nous Research models](https://docs.oceantoken.ai/models/nousresearch): 3 models - [NVIDIA models](https://docs.oceantoken.ai/models/nvidia): 15 models - [OpenAI models](https://docs.oceantoken.ai/models/openai): 80 models - [Perceptron models](https://docs.oceantoken.ai/models/perceptron): 1 models - [Perplexity models](https://docs.oceantoken.ai/models/perplexity): 7 models - [Poolside models](https://docs.oceantoken.ai/models/poolside): 4 models - [Qwen models](https://docs.oceantoken.ai/models/qwen): 49 models - [Recraft models](https://docs.oceantoken.ai/models/recraft): 15 models - [Rekaai models](https://docs.oceantoken.ai/models/rekaai): 2 models - [Relace models](https://docs.oceantoken.ai/models/relace): 2 models - [Runway models](https://docs.oceantoken.ai/models/runway): 2 models - [Sakana AI models](https://docs.oceantoken.ai/models/sakana): 3 models - [Sao10k models](https://docs.oceantoken.ai/models/sao10k): 3 models - [Sentence Transformers models](https://docs.oceantoken.ai/models/sentence-transformers): 5 models - [Sesame models](https://docs.oceantoken.ai/models/sesame): 1 models - [Sourceful models](https://docs.oceantoken.ai/models/sourceful): 4 models - [StepFun models](https://docs.oceantoken.ai/models/stepfun): 2 models - [Tencent models](https://docs.oceantoken.ai/models/tencent): 7 models - [Thedrummer models](https://docs.oceantoken.ai/models/thedrummer): 3 models - [Thenlper models](https://docs.oceantoken.ai/models/thenlper): 2 models - [Thinkingmachines models](https://docs.oceantoken.ai/models/thinkingmachines): 2 models - [Typesafe models](https://docs.oceantoken.ai/models/typesafe): 1 models - [Undi95 models](https://docs.oceantoken.ai/models/undi95): 1 models - [Upstage models](https://docs.oceantoken.ai/models/upstage): 2 models - [Voyage AI models](https://docs.oceantoken.ai/models/voyageai): 7 models - [Writer models](https://docs.oceantoken.ai/models/writer): 1 models - [xAI models](https://docs.oceantoken.ai/models/x-ai): 13 models - [Xiaomi models](https://docs.oceantoken.ai/models/xiaomi): 2 models - [Z.ai models](https://docs.oceantoken.ai/models/z-ai): 16 models ## Machine-readable - [llms-full.txt](https://docs.oceantoken.ai/llms-full.txt): this whole documentation as one Markdown file, every model and every endpoint included - [Public model catalog (JSON)](https://api.oceantoken.ai/public/model_hub): the live catalog the pages are built from - [Agent install guide](https://mcp.oceantoken.ai/install): how an AI agent installs OceanToken into Codex, Claude Code, Cursor, Gemini CLI or OpenClaw ## Optional - [Product site](https://oceantoken.ai) - [Console](https://app.oceantoken.ai/ui/login/) - [Plugins for coding agents (GitHub)](https://github.com/NextFormAI/oceantoken-plugins)