MINIMAX / Prompt enhancement
H3 Context-IR
minimax-h3-context-ir
Enhance a creative prompt and receive text for the next generation step.
- Output
- Enhanced prompt text
- Resolution
- Not applicable
- Target video pacing (text output)
- 4–15 seconds
- Integration
- HTTP / asynchronous tasks
CAPABILITIES
Capabilities & limits
Supported modes
Before you use this model
The result is text, not video. Duration describes the intended video pacing and does not cap output tokens. Input and output tokens are charged separately. Text-only enhancement requires an explicit aspect ratio.
| Parameter | Supported values / default |
|---|---|
| Resolution | Not applicable |
| Target video pacing (text output) | 4–15 seconds |
| Aspect ratio | 21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16 · adaptive |
| Prompt limit | 7000 Unicode code points |
| Definition defaults (subject to mode rules) | duration: 5 · aspect_ratio: 16:9 |
| Reference media limits | 9 images / 3 videos / 3 audio clips |
Media limits do not mean all inputs can be combined in every mode. Consult the full field reference for sizes, formats and incompatible combinations.
Explore the models supported by ByteSOTA. Availability depends on your account permissions; check /v1/models for enabled models, configurations and defaults.
Full request fields and limits (Chinese) ↗CREATIVE WORKFLOWS
Use cases
- Develop a creative idea into a richer prompt.
- Prepare directions before a separate video generation request.
- Use reference material to guide prompt development.
No verified public output sample is available for this model yet. The request example below comes from the public integration contract and illustrates the API only.
PRICING / USD
Price the specification you need.
Prices are in USD, with a date shown for official reference rates. Missing prices do not mean free usage. Existing tasks are billed using the prices recorded when they were created.
Loading current prices…
API INTEGRATION
API example
Send HTTP requests from your server using the model and parameters you choose. ByteSOTA must first enable your account, API key and model permissions.
HTTP 202 means accepted. Save the task ID and follow next_poll_after_seconds. Generation status, result availability and billing status are separate facts.
curl --request POST 'https://bytesota-api.businessmodelcraft.com/v1/videos/generations' \
--header 'Authorization: Bearer YOUR_PLATFORM_API_KEY' \
--header 'Content-Type: application/json' \
--header 'Idempotency-Key: example-ir-text-001' \
--data-raw '{
"model": "minimax-h3-context-ir",
"prompt": "A kitten walks along a garden path as the camera follows slowly.",
"duration": 5,
"aspect_ratio": "16:9"
}'
Query result
curl 'https://bytesota-api.businessmodelcraft.com/v1/tasks/vid_ABCDEFGHIJKLMNOP' \
--header 'Authorization: Bearer YOUR_PLATFORM_API_KEY'
Copy this example into your project. It does not run on this page, and the API key shown is a placeholder.
Specifications and examples: model definitions and the public API integration guide.
Read the full API guide (Chinese) ↗