Doubao Vision Models by ByteDance
Last refreshed 2026-10-17. Next refresh: weekly.
Details
Capabilities
Links
WebsiteAbout
Doubao Vision is ByteDance’s Doubao visual-understanding family of multimodal language models (VLMs)—built to analyze images and related visual inputs and answer in text, not to generate images or video. The Lite and Pro 32K-class models Doubao-vision-lite-32k and Doubao-vision-pro-32k are offered through Volcengine Ark as conversation and multimodal models; first-party Ark developer material presents Doubao-vision-pro-32k as a VLM that reads visual information in pictures for tasks such as describing an image. Later Doubao 1.5 vision-understanding releases on Ark continue this understanding line, documenting upgrades in visual reasoning, document and text recognition, information extraction and OCR, and image and video understanding via the Chat Completions visual-understanding API.
Decision facts
- Best fit
- vision and multimodal work
- Capability starting point
- Doubao Vision Lite 32K with 32K context and multimodal inputs
- Lowest tracked input
- Not tracked
- Closest related family
- BAGEL
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
Use when the workload needs 32K context, 32B parameters, and multimodal inputs.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Doubao Vision Lite 32K | Use when the workload needs 32K context, 32B parameters, and multimodal inputs. | 2024-12 | 32K context32B parametersmultimodal inputs | Current |
Release Timeline
1 release groupSpecifications(2 models)
| Model | Released | Context | Parameters | Vision |
|---|---|---|---|---|
| Doubao Vision Lite 32K | 2024-12 | 32K | 32B | Yes |
