moonshot-v1-vision-preview

Kimi visual model

Kimi visual model is an advanced image understanding technology provided by the Moonshot AI open platform, which can accurately identify and understand image content, including text, color, object shape, etc. It is efficient and accurate and suitable for various scenarios such as image content description and visual question and answer. The pricing is consistent with the moonshot-v1 series model. It is billed based on the total Tokens inferred by the model, and each picture consumes 1024 Tokens.

target users

Developers, researchers, and businesses requiring image understanding capabilities. Developers can easily integrate its powerful API interface; researchers can use it for image analysis and research; enterprises can improve business efficiency and user experience.

Usage scenario examples

Developer develops image question and answer application

Enterprises conduct automated image content review

Researchers conduct image recognition study

Product features

Support multiple rounds of conversations, understand context and answer questions

Provide streaming output and return results in real time

Tool calls can be made to expand the scope of application

Support JSON mode to facilitate data interaction

Support partial processing and response to improve efficiency

Internet search is not supported to ensure data security

Creating caches with image content is not supported, but already created caches can be used

Only supports base64 encoded image content

Tutorial

1 Get the Moonshot API key

2 Select the appropriate Kimi vision model, such as moonshot-v1-8k-vision-preview

3 Convert the image to base64 encoded string

4 Build an API request, including model name, image content and instructions

5 Send a request to the Moonshot AI open platform

6 Parse the response results and perform subsequent processing