Model selection & comparison
Choose the right Kimi API model based on capability, speed, and pricing.
Kimi API offers multiple models for developers to choose from.
Available models
Visit platform.kimi.ai/docs/models for the complete model list and detailed specifications. When selecting a model, you can consider:
| Dimension | Consideration |
|---|---|
| Context Length | Choose large-context models for long document processing |
| Response Speed | Lightweight models for latency-sensitive scenarios |
| Generation Quality | Higher-tier models for complex reasoning and creative writing |
| Pricing | Choose cost-effective model based on budget and volume |
Vision models
Vision models support image input for captioning, OCR, chart interpretation, and more:
- Each image is billed at a flat rate of 1,024 tokens, regardless of dimensions or resolution
- Supported formats: JPEG, PNG, WebP, etc.
- Images can be passed via URL or Base64 encoding
Currently unsupported capabilities
- Video Multimodal: Direct video input not yet supported
- PPT Generation API: Not yet available via API
- Deep Research API: Not yet available via API
For updates on these capabilities, follow the platform announcements.