Below are frequently asked questions about Kimi API model capabilities.

Support JSON mode?

Yes. Kimi API provides JSON Mode, which constrains the model output to valid JSON. Usage:

• Set the response_format parameter to {"type": "json_object"} in your request.
• Also explicitly instruct the model in your prompt to return JSON and describe the expected structure.
• In JSON Mode, the model is guaranteed to output a parseable JSON string.

Support model fine-tuning?

Kimi API does not currently support user-initiated model fine-tuning. For custom model requirements, contact the sales team via platform.moonshot.ai/contact-sales to learn about enterprise customization options.

Support OCR?

Kimi API's Vision models have image understanding capabilities and can recognize text within images, achieving OCR-like results:

• Supports recognition of text in English, Chinese, and many other languages.
• Can recognize printed text and some handwritten text.
• Each image is billed at a flat rate of 1,024 tokens.
• Beyond plain text recognition, the models can also interpret tables, charts, and other structured content within images.

Support TTS (text-to-speech) and ASR (speech recognition) ?

Kimi API does not currently support TTS or ASR. For speech-related capabilities, consider combining a third-party speech service with Kimi API. Follow platform announcements for future feature updates.