A Shift in Billing: From Tokens to Requests
On April 9, 2026, iFlytek officially implemented a major upgrade for its “iFlytek Coding Plan” service, fundamentally changing its billing model. The service has moved away from the industry-standard, often unpredictably costly, token-based billing in favor of a request-based subscription model. This adjustment is designed to offer developers a more stable and predictable cost structure, eliminating the “token anxiety” that arises from high token consumption when handling long texts or complex tasks.
The upgraded Coding Plan offers three tiered packages to meet different development needs:
- Starter Plan: Priced at ¥3.9 for the first month, offering unlimited requests. This plan mainly supports lightweight models, including
Qwen3.5-35B-A3B, Qwen3-Coder-Next-FP8, and GLM-4.7-Flash, suitable for batch processing, general-purpose chat, or simple coding tasks.
- Pro Plan: Priced at ¥39/month, this plan is positioned as a high-value option. The default model is
GLM-5, with support for other models like Kimi-K2.5, MiniMax-M2.5, and DeepSeek-V3.2. The request quota is 18,000 per month, with rate limits of 9,000 per week and 1,200 per 5 hours. It is suitable for moderate-intensity programming assistance and AI Agent development.
- Performance Plan: Priced at ¥199/month and designed for heavy usage. This plan supports the latest
GLM-5.1 model and provides a high quota of up to 90,000 requests per month (rate-limited to 45,000 per week and 6,000 per 5 hours). It is ideal for parallel development on multiple projects and complex software engineering tasks.
It is worth noting that the platform indicates the calculation of request counts may be related to task complexity, a common mechanism in similar API services.
Core Model Update: Integrating the Open-Source Leader, GLM-5.1
One of the main highlights of this Coding Plan upgrade is the rapid integration of the GLM-5.1 model, which was officially open-sourced by Zhipu AI on April 8, 2026. Released under the MIT License, this model achieved a score of 58.4 on the authoritative software engineering benchmark, SWE-Bench Pro.
According to the published data, GLM-5.1's score surpasses several leading closed-source models from the same period, such as Claude Opus 4.6 (57.3), GPT-5.4 (57.7), and Gemini 3.1 Pro (55.1). This marks the first time an open-source model has achieved the top score on this benchmark, which focuses on evaluating a model’s ability to solve real-world programming problems. iFlytek’s integration of GLM-5.1 just one day after its open-source release demonstrates the company’s agility in responding to cutting-edge developments in the open-source community.
Developer Integration Guide: Obtaining Your API Key
To use the iFlytek Coding Plan, developers first need to obtain API credentials through the iFlytek MaaS platform. Here are the steps:
- Visit the official iFlytek MaaS platform website and log in.
- On the platform’s homepage, find the “Coding Plan Upgrade” entry point and navigate to the subscription page.
- Select the “Starter”, “Pro”, or “Performance” plan based on your needs and complete the payment.
- After subscribing successfully, go to the “CodingPlan” management page from the left-hand navigation bar.
- In the “Key Configuration Information” section of the details page, you can find and copy the auto-generated API Key. This key serves as the unique credential for all subsequent API calls.
Additionally, the platform provides a unified model ID for all calls: astron-code-latest. Developers can switch the underlying model bound to their API key by selecting “Configure Model” from the “Actions” menu on the plan details page.
Practical Application: Configuring Mainstream Development Tools
_iFlytek Coding Plan simplifies integration with various development tools by providing API endpoints compatible with mainstream protocols.
Integrating with Claude Code
Claude Code is a local AI coding assistant. Developers can integrate it with the iFlytek Coding Plan by following these steps:
- Install Claude Code: If not already installed, you can use the official one-click installation script. For macOS/Linux/WSL, use
curl -fsSL https://claude.ai/install.sh | bash; for Windows PowerShell, use irm https://claude.ai/install.ps1 | iex.
- Configuration: Edit the Claude Code configuration file at
~/.claude/settings.json, add the following content, and replace the value of ANTHROPIC_AUTH_TOKEN with your own API Key.
{
“env”: {
“ANTHROPIC_AUTH_TOKEN”: “Your API Key”,
“ANTHROPIC_BASE_URL”: “https://maas-coding-api.cn-huabei-1.xf-yun.com/anthropic”,
“CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC”: 1,
“API_TIMEOUT_MS”: 600000,
“ANTHROPIC_MODEL”: “astron-code-latest”,
“ANTHROPIC_SMALL_FAST_MODEL”: “astron-code-latest”
},
“permissions”: {
“allow”: [],
“deny”: []
}
}
This configuration uses the Anthropic-compatible API endpoint provided by iFlytek. If you are prompted to log in on the first launch, you can skip it by adding the "hasCompletedOnboarding": true field to the configuration file.
Integrating with OpenClaw and Other OpenAI-Compatible Tools
For tools that support the OpenAI API format, such as the AI Agent framework OpenClaw or the editor Cursor, you can use the OpenAI-compatible endpoint provided by iFlytek for configuration.
Taking OpenClaw as an example, the configuration steps are as follows:
- Edit Configuration File: Open the
~/.openclaw/openclaw.json file.
- Add Model Provider: In the
models.providers section, add the configuration for iFlytek Coding Plan, and set it as the primary model in agents.defaults. Be sure to replace apiKey with your own key.
{
“models”: {
“mode”: “merge”,
“providers”: {
“astroncodingplan”: {
“baseUrl”: “https://maas-coding-api.cn-huabei-1.xf-yun.com/v2”,
“apiKey”: “Your API Key”,
“api”: “openai-completions”,
“models”: [
{
“id”: “astron-code-latest”,
“name”: “astron-code-latest”,
“reasoning”: false,
“input”: [“text”],
“contextWindow”: 200000,
“maxTokens”: 128000
}
]
}
}
},
“agents”: {
“defaults”: {
“model”: {
“primary”: “astroncodingplan/astron-code-latest”
},
“models”: {
“astroncodingplan/astron-code-latest”: {
“alias”: “astroncodingplan”
}
}
}
}
}
- Restart Service: After saving the configuration, run
openclaw gateway restart in your terminal to apply the changes. Once configured, you can start using the model service provided by iFlytek Coding Plan within OpenClaw.
