Mini Program Growth Plan User Guide
Use Text Generation Models
The tokens gifted by the Growth Plan can be used to call text generation models directly in your mini program. Choose the SDK for your runtime environment:
- Mini Program: Mini Program Integration
- Cloud Functions: wx-server-sdk Integration
- Cloud Run / Node.js: Node SDK Integration
The Token gifted by the Growth Plan currently covers six models: hy4-flash, hy4, hy4-preview, hy3, deepseek-v4.1-flash, and deepseek-v4.1-flash-1017. Calling other models does not consume this quota.
Code example:
const model = ai.createModel("cloudbase");
const res = await model.invoke({
model: "deepseek-v4.1-flash",
messages: [{ role: "user", content: "Hello" }],
});
cloudbase is the recommended model group. The existing hunyuan-v3 group differs as follows:
| Aspect | cloudbase | hunyuan-v3 |
|---|---|---|
| Supported Models | All models | hy3 only |
| Free Quota Consumption | Free quota is consumed first when the source is allowed | Only the free quota is consumed |
| When Free Quota Exhausted | Automatically falls back to plan quota | Returns an error |
| When Source Not Allowed | Automatically falls back to plan quota | Returns an error |
| Plan Quota | Supported | Not supported |
| Applicable Plan | Resource-Point plan only | Available on both Resource-Point and non-Resource-Point plans |
| Model Toggle | No enablement required; can be disabled | No enablement required; cannot be disabled |
| Roadmap | New models continue to be added | No new models; being deprecated. Migration guide |
The free AI Resource Pack quota from the "Mini Program Growth Plan" is only for Mini Programs and the CloudBase server-side, and currently covers hy4-flash, hy4, hy4-preview, hy3, deepseek-v4.1-flash, and deepseek-v4.1-flash-1017. When using AI tools or other non-Mini Program scenarios, calling these models will be deducted from your plan quota instead.
See the Special Governance Announcement for usage restrictions and compliance requirements.
The hy4-preview model will be taken offline on December 7, 2026. Access to this model will no longer be available after that time. Please switch the model parameter in your code to hy4. See Notice on the Offline of the Hy4 Preview Model.
Use Image Generation Models
The image generation credits gifted by the Mini Program Growth Plan can be used to call hy-image-v3.5-preview, which generates images from text or a reference image. You can call it from a Mini Program or from the server side. For parameters and the response structure, see Hy-Image-3.5-preview.
- Mini Program: Mini Program
- Cloud Functions: wx-server-sdk
- Cloud Run / Node.js: Node SDK
A well-crafted prompt significantly improves the quality of generated images:
- Describe the subject: "an orange cat" → "a chubby, adorable orange cat with big eyes and short legs"
- Specify a style: add descriptors like "watercolor style", "oil painting style", "cyberpunk", or "realistic photography"
- Describe the scene and lighting: "a sun-drenched café", "golden hour soft side lighting"
- Specify the composition: "close-up", "bird's-eye view", "wide shot", "center-symmetric composition"
Example:
A chubby orange cat napping on a wooden windowsill in the sunshine, warm sunlight on its fluffy fur, watercolor illustration style, soft warm tones, cozy and heartwarming scene
Code example:
const imageModel = ai.createImageModel("cloudbase");
// Text-to-image
const res1 = await imageModel.generateImage({
model: "hy-image-v3.5-preview",
messages: [
{ role: "user", content: [{ type: "text", text: "A cat on the grass" }] },
],
});
// Reference-image generation: mix text and images in content
const res2 = await imageModel.generateImage({
model: "hy-image-v3.5-preview",
messages: [
{
role: "user",
content: [
{ type: "text", text: "Draw a cat in the style of this image, oil painting" },
{ type: "image_url", image_url: { url: "https://example.com/reference.jpg" } },
],
},
],
});
HY-Image-3.0-Plus-4090-Tob-v1.0 and HY-Image-v3.0-I2I-ToB-v1.0.1 will be taken offline at 00:00 on December 1, 2026, and calls will fail. For the parameters and response structure of the legacy models, see Legacy Models.
Develop Mini Programs with AI Tools
Use CloudBase Skills (MCP) to let Cursor, Lobechat 🦞, OpenClaw, CodeBuddy, WorkBuddy, VS Code, Claude Code, and other AI tools directly operate CloudBase resources without manual environment setup.
Step 1, run the following command in your AI tool to install CloudBase Skills:
npx skills add tencentcloudbase/cloudbase-skills -y
Step 2, tell your AI: "Use CloudBase Skills to develop a todo mini program"
The AI will automatically handle: code generation, database creation, cloud function deployment, and more.
For detailed configuration, see CloudBase MCP/Skill Configuration Guide.
Get Help
Visit the CloudBase Community to ask questions, join communication groups, or submit tickets — 1v1 dedicated customer service available.