ByteDance: Doubao-vision-lite-32k API

A cost-optimized multimodal model featuring a 32k context window and sparse MoE architecture. It balances rapid visual recognition and text reasoning, ideal for high-throughput tasks like batch OCR and UI element detection.

Frequently Asked Questions

What is the context window of Doubao-vision-lite-32k?

Doubao-vision-lite-32k supports a context window of up to 32,768 tokens.

Does Doubao-vision-lite-32k support function calling?

Yes. Doubao-vision-lite-32k supports tool / function calling.

What is the knowledge cutoff of Doubao-vision-lite-32k?

The knowledge cutoff of Doubao-vision-lite-32k is 2025-08-31.