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.
- Context window: 32,768 tokens
- Max output: 4,096 tokens
- Input: text, image
- Output: text
- Tool calling: Supported
- File input: Supported
- Released: 2025-01-21
- Knowledge cutoff: 2025-08-31
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.