Get started with Token Plan Individual in three steps.
Step 1: Subscribe to Token Plan Individual
Go to the Token Plan Individual purchase page, select a plan and subscription period, and complete the purchase.
Step 2: Get your API Key and Base URL
- API Key: After subscribing, generate your API Key from the Token Plan console. The API Key is displayed in full only once -- copy and save it immediately.
- Base URL: Choose the Base URL that matches the protocol your AI tool supports.
| Protocol | Base URL |
|---|---|
| OpenAI compatible | https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 |
| Anthropic compatible | https://token-plan.ap-southeast-1.maas.aliyuncs.com/apps/anthropic |
Token Plan API Keys start with
sk-sp- and are different from regular QwenCloud API Keys (sk-ws- prefix). Token Plan, Coding Plan, and pay-as-you-go API Keys and Base URLs are completely separate and must be used together. See the difference between API key types for details.Step 3: Set up your AI tool
Configure the API Key and Base URL in your AI tool to start using the service.
OpenClaw
Open-source, self-hosted personal AI assistant
Hermes Agent
Open-source AI agent framework with built-in self-learning loop
Claude Code
AI terminal coding assistant with natural language programming
OpenCode
Open-source AI coding agent tool
Cursor
AI-native code editor
Codex
Command-line coding tool by OpenAI
Qwen Code
Open-source command-line AI coding tool
QwenPaw
Open-source personal AI assistant for local and cloud deployment
Cherry Studio
Multi-model desktop client
Chatbox
Cross-platform AI desktop client
Cline
VS Code extension for intelligent code completion and debugging
Qoder
Agentic coding platform for real-world software development
Lingma
Alibaba Cloud intelligent coding assistant with standalone IDE
Kilo CLI
Lightweight, high-performance command-line coding tool
More Tools
Other coding tools
Optional: Integrate multimodal generation models
Token Plan Individual supports multimodal generation models (wan2.7-image, happyhorse-1.1-t2v, etc.). Multimodal generation models are not compatible with the standard text generation endpoint (/api/v1/services/aigc/text-generation/generation) or the OpenAI-compatible endpoint (/compatible-mode/v1/chat/completions). You must call them through their dedicated endpoints. For example, video generation models such as happyhorse-1.1-t2v use the /api/v1/services/aigc/video-generation/video-synthesis endpoint. Integrate them through the Skill or extension mechanism of your AI tool. For details, see Integrate multimodal generation models.
FAQ
What should I do if calling a multimodal generation model returns a url error or a model_not_supported error?
Multimodal generation models do not support the standard text generation endpoint or the OpenAI-compatible endpoint. Calling them through the wrong endpoint returns an InvalidParameter error with the message "url error", or a model_not_supported error. Switch to the dedicated endpoint: video generation models use /api/v1/services/aigc/video-generation/video-synthesis, and image generation models use /api/v1/services/aigc/image-generation/generation.
Optional: Integrate Harness tools
Some Qwen models (qwen3.7 and qwen3.8 series) have built-in Harness tools that extend capabilities such as web search, text-to-image search, reverse image search, web scraping, and code interpreter during conversations. Harness tools can only be invoked through the Responses API, and Credits are deducted based on the number of successful invocations. For details, see Integrate Harness tools.
Harness tools must be called via the Responses API to be triggered automatically. If your AI tool only supports the Chat Completions protocol, the model will respond normally, but will not automatically invoke Harness tools — such requests will be billed on a pay-as-you-go basis and will not consume Harness tool Credits from your plan quota. To use Harness tools, choose an AI tool that is compatible with the Responses API (see Integrate Harness tools).