TokenSea
TokenSea Integration Hub

OpenAI-compatible integration docs for coding agents, SDKs, and AI clients.

TokenSea provides one OpenAI-compatible endpoint for accessing supported AI models. Use your TokenSea API key with SDKs, coding agents, agent platforms, and OpenAI-compatible clients.

Create API Key
Global constants
Base URL https://api.tokensea.world/v1
API key format sk_live_xxxxxxxxxxxxxxxxxxxx
Example models deepseek-v4-flash deepseek-v4-pro gpt-5.6-sol (provider setup required)
Compatibility note Most tools only need API key, Base URL, and Model ID.

Quick Start

TokenSea gives you an OpenAI-compatible endpoint for chat clients, coding agents, and SDKs. Create an account with any email, add balance with PayPal, create an API key, then paste the key and Base URL into your tool.

Step 1. Sign up

Create an account at tokensea.world with any email address. Verify the email if prompted, then open your dashboard.

Step 2. Add balance

Use the billing page on tokensea.world and pay via PayPal. Your available balance is shown in the dashboard after payment is confirmed.

Step 3. Create a key

Open Dashboard > API Keys, create a key, and keep the full value somewhere secure. The key is only needed in clients that make API calls.

Step 4. Configure a client

Use Base URL https://api.tokensea.world/v1, your TokenSea API key, and a currently available model ID such as deepseek-v4-flash.

curl https://api.tokensea.world/v1/chat/completions \
  -H "Authorization: Bearer YOUR_TOKENSEA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {"role": "user", "content": "Reply with PONG"}
    ],
    "stream": false
  }'

API Keys

API keys authorize requests to TokenSea. Create them from the API Keys page in your dashboard and paste them into clients as a Bearer token or OpenAI API key.

Create and copy

  1. Open Dashboard > API Keys.
  2. Click the create key action.
  3. Name the key for the client or environment using it.
  4. Copy the full sk-... value before closing the dialog.

Operate safely

  • Use separate keys for production, local testing, and each shared workstation.
  • Delete and recreate a key immediately if it appears in logs, screenshots, or chat history.
  • Do not add IP or model restrictions while first testing a new client.

Base URL

Most clients need one endpoint, one key, and one model. TokenSea uses the standard OpenAI-compatible /v1 API path.

Production

https://api.tokensea.world/v1

Use this value for deployed clients and normal user traffic.

Local gateway

http://localhost:8001/v1

Use this value only when you are running the TokenSea gateway locally.

Some clients split the host and path. In that case use host https://api.tokensea.world and path /v1. If a field says API Host and warns not to include a trailing slash, use https://api.tokensea.world.

Models & Pricing

Pass the exact model ID returned by your TokenSea key. DeepSeek V4 is live now. OpenAI's current GPT-5.6 Sol, Terra, and Luna IDs are tracked here, but remain unavailable until the GPT provider is connected.

deepseek-v4-flash · available deepseek-v4-pro · available gpt-5.6-sol · setup required gpt-5.6-terra · setup required gpt-5.6-luna · setup required

How to list models

curl https://api.tokensea.world/v1/models \
  -H "Authorization: Bearer YOUR_TOKENSEA_API_KEY"

Pricing

Prices vary by model and by input, output, and cached token type. Check tokensea.world before large runs, because routing and upstream availability can change.

Billing Basics

TokenSea uses account balance. Add balance with PayPal, then API usage draws down that balance based on the model used and the tokens processed.

Token billing

Tokens are the units models process. English words, punctuation, code, JSON, and non-English characters all become tokens before billing.

Cost = input tokens * input price
     + output tokens * output price
     + cached token charges when applicable

Practical controls

  • Use lower-cost models for drafts, tests, and batch jobs.
  • Keep prompts concise when the task does not need long context.
  • Use streaming for better UX, but remember billing is based on tokens processed.
  • Check usage logs when a client retries automatically.

FAQ

Common TokenSea configuration and runtime issues.

Invalid API key

Confirm the key starts with sk-, has no extra spaces, and is in the API key field rather than the model field. If unsure, create a new key from Dashboard > API Keys.

Model not available

Run GET /v1/models and copy the exact model ID. If a client cannot fetch models, add deepseek-v4-flash manually.

Balance errors

Add balance via PayPal on tokensea.world and check Dashboard > Usage. Some clients retry failed requests, so repeated failures can create multiple log entries.

Short or empty output

Reduce prompt length, lower concurrency, try a smaller response limit, or switch models. Very long contexts and upstream load can cause incomplete responses in some clients.

socket hang up

This usually means the client connection was interrupted. Check local network stability, proxy settings, and whether the client is using the full Base URL with /v1.

Invalid parameters

Use a simple chat request first. Remove unsupported custom parameters, confirm the selected model is a chat-capable model, and update the client if it sends outdated OpenAI fields.

CherryStudio

CherryStudio is a desktop AI chat client. Configure TokenSea as an OpenAI-compatible provider.

  1. Open CherryStudio and click the settings icon in the lower-left corner.
  2. Open Model Service, then click + Add.
  3. Set provider name to TokenSea and provider type to OpenAI.
  4. Set API key to your TokenSea key.
  5. Set API address to https://api.tokensea.world if the field asks for host only, or https://api.tokensea.world/v1 if it asks for Base URL.
  6. Click Manage, add deepseek-v4-flash, then run Check.
Do not leave an old default model selected. If the check fails, manually type an available model ID from the Models & Pricing section.

Chatbox

Chatbox supports custom OpenAI-compatible providers on desktop and mobile.

  1. Click the settings gear, open Model Provider, scroll to the bottom, and click + Add.
  2. Set Name to TokenSea.
  3. Set API Mode to OpenAI API Compatible.
  4. Set API Key to your TokenSea key.
  5. Set API Host to https://api.tokensea.world with no trailing slash.
  6. Click Fetch. If model fetching does not work, click + New and add deepseek-v4-flash manually.
  7. Click Check, select the model, and confirm the connection.

ChatGPT Next Web

ChatGPT Next Web can use TokenSea through its custom interface settings.

  1. Open ChatGPT Next Web and go to settings.
  2. Find the custom endpoint or custom interface section.
  3. Set interface address to https://api.tokensea.world or the full Base URL if your deployment expects /v1.
  4. Paste your TokenSea API key.
  5. Select or type a model ID such as deepseek-v4-flash.
  6. Return to the chat page and send a short test message.
Avoid a trailing slash after the endpoint. If the app appends /v1 automatically, use the host-only value.

LobeChat

LobeChat supports custom OpenAI-compatible model providers.

  1. Open settings and go to the language model provider settings.
  2. Choose the OpenAI provider or add a custom OpenAI-compatible provider.
  3. Set API proxy address or Base URL to https://api.tokensea.world/v1.
  4. Paste your TokenSea API key.
  5. Add deepseek-v4-flash manually if needed.
  6. Use the built-in check action before starting a long conversation.

SillyTavern

Use SillyTavern as an advanced chat front end with TokenSea's OpenAI-compatible chat completions endpoint.

  1. Open the API connection panel.
  2. Create a new API connection profile.
  3. Set API type to Chat Completion.
  4. Set source to Custom (OpenAI-compatible).
  5. Set custom endpoint/Base URL to https://api.tokensea.world/v1.
  6. Set custom API key to your TokenSea key.
  7. Set model name to deepseek-v4-flash or another available chat model.
  8. Click Connect and verify that the status becomes valid.
If a response is cut short, reduce context size, disable aggressive concurrent requests, or switch to a lower-latency model.

Tavo

Tavo can connect to TokenSea when using its custom OpenAI protocol option.

  1. Open the side menu and go to API Connection.
  2. Click + New.
  3. Choose Custom (OpenAI protocol).
  4. Set API endpoint to https://api.tokensea.world/v1.
  5. Paste your TokenSea API key.
  6. Choose or manually type deepseek-v4-flash.
  7. Save the profile and send a short test message.

OpenCat

OpenCat is a native macOS and iOS chat client with custom API service support.

  1. Open settings in OpenCat.
  2. Add a new API service or edit an existing custom OpenAI service.
  3. Set API domain to https://api.tokensea.world.
  4. Paste your TokenSea API key.
  5. Add the model ID deepseek-v4-flash.
  6. Use Verify or the connection test action, then return to chat.
If OpenCat exposes a full Base URL field instead of API domain, include /v1.

BotGem

BotGem is a cross-platform chat client. Configure TokenSea through OpenAI-compatible mode.

  1. Open BotGem and go to settings.
  2. Open API configuration.
  3. Choose OpenAI-compatible mode.
  4. Set API Host to https://api.tokensea.world.
  5. Paste your TokenSea API key.
  6. Save settings, return to chat, and select an available model.

ccswitch

ccswitch manages provider profiles for coding tools such as Claude Code, Codex, Gemini CLI, OpenCode, and OpenClaw. Add TokenSea once, then switch tools to the TokenSea profile where supported.

Install

  1. Download the installer from the ccswitch release page.
  2. Install the app. On Windows, use More info and Run anyway if SmartScreen appears.
  3. Open settings and enable the plugin takeover switch for the tools you want ccswitch to manage.

Add TokenSea

  1. Click the + action to add a provider.
  2. Use provider name TokenSea.
  3. Set API key to your TokenSea key.
  4. Use https://api.tokensea.world/v1 for OpenAI-compatible request address fields.
  5. Set the default model to deepseek-v4-flash.
If ccswitch says a Universal Provider field auto-appends paths, use host https://api.tokensea.world there. Restart the terminal session after switching providers.

Cursor

Cursor can use TokenSea for features that honor custom OpenAI-compatible settings.

  1. Open Cursor Settings.
  2. Open the Models tab.
  3. Find the OpenAI API Key section.
  4. Set Override OpenAI Base URL to https://api.tokensea.world/v1.
  5. Set OpenAI API Key to your TokenSea key.
  6. Click Verify, then choose deepseek-v4-flash.
  7. Restart Cursor if the old provider remains active.
Some Cursor features may still use Cursor's native provider stack. Test chat and composer behavior separately.

VS Code Copilot

VS Code extension support for custom API endpoints depends on the exact extension and version. Use TokenSea only where custom OpenAI-compatible settings are exposed.

  1. Install or update the VS Code AI extension you plan to use.
  2. Open VS Code settings with Ctrl + ,.
  3. Search for provider, proxy, OpenAI, or custom API settings.
  4. Set API address/Base URL to https://api.tokensea.world/v1.
  5. Paste your TokenSea API key.
  6. Set the model to deepseek-v4-flash for general work or deepseek-v4-pro for stronger reasoning.
  7. Reload VS Code after saving settings.
If an extension only supports the native GitHub Copilot service and does not expose a custom endpoint, it cannot be pointed at TokenSea directly.

Cline

Cline is a VS Code coding assistant extension. Select its OpenAI-compatible provider mode for TokenSea.

  1. Install Cline in VS Code.
  2. Open the Cline sidebar and click the settings icon.
  3. Set API Provider to OpenAI Compatible.
  4. Set Base URL to https://api.tokensea.world/v1.
  5. Set API Key to your TokenSea key.
  6. Set the model to deepseek-v4-flash or another currently available model.
  7. Start with a small request before running repository-wide edits.

OpenCode

OpenCode can be configured directly or through ccswitch. The ccswitch path is usually easier if you also switch providers for other coding tools.

Install OpenCode

curl -fsSL https://opencode.ai/install | bash
npm i -g opencode-ai
bun add -g opencode-ai
brew install anomalyco/tap/opencode

Configure via ccswitch

  1. Open ccswitch and switch to the OpenCode tab.
  2. Click Add Provider.
  3. Choose custom configuration.
  4. Set provider ID to tokensea and format to OpenAI Compatible.
  5. Set Base URL to https://api.tokensea.world/v1.
  6. Add option setCacheKey = true if the UI exposes extra options.
  7. Select deepseek-v4-flash, save, and add the provider to OpenCode.
{
  "provider": {
    "tokensea": {
      "options": {
        "apiKey": "YOUR_TOKENSEA_API_KEY",
        "baseURL": "https://api.tokensea.world/v1"
      }
    }
  },
  "model": "deepseek-v4-flash"
}
Restart an already-running OpenCode terminal after ccswitch writes a new provider profile.