Supported Models and ParametersText ModelsDeepSeek
DeepSeek V4 Flash Vision
DeepSeek V4 Flash Vision text model parameters and pricing.
deepseek_v4_flash_vision_exp is a currently public China-region text model for multimodal text generation with reference images.
Actual parameters must follow the schema.fields returned by GET /v1/models.
Quick Facts
| Item | Value |
|---|---|
| Model ID | deepseek_v4_flash_vision_exp |
| Type | Text |
| Region | CN |
| Best For | multimodal text generation with reference images |
Recommended Request Shape
Use /v1/chat/completions or /v1/responses for text models, not the media generation endpoints.
{"model":"deepseek_v4_flash_vision_exp","input":"Write a launch announcement","stream":false}Field Reference
| Field | Type | Required | Description |
|---|---|---|---|
model | string | Yes | Must be deepseek_v4_flash_vision_exp |
input | string | Yes | User input content |
reference_images | string[] | No | Reference images, up to 5; only send when declared by the live schema |
stream | boolean | Schema-dependent | Whether to stream |
max_output_tokens | integer | Schema-dependent | Maximum output tokens |
Current Public Price
| Charge | Credits / 1M tokens |
|---|---|
| Input | 120 |
| Output | 360 |
| Cache read/write | 0 / 0 |
Parameters and pricing
- Do not send
reasoning,temperature, or other fields unless the model schema declares them. - Use
/v1/quoteas the final pricing authority.
Smallest Example
{"model":"deepseek_v4_flash_vision_exp","input":"Write a launch announcement"}