China78B parametersImage + videoMultilingual

Explore InternVL3-78B

Use OpenGVLab's large multimodal model for multi-image dialogue, charts, documents, spatial reasoning, and video understanding.

Model

OpenGVLab/InternVL3-78B

Input

Configure your request

System message Optional

Question or instruction

Images Optional

Upload one or several images.

Video Optional

Optional video input sampled into frames.

Conversation history (JSON) Optional

Parameters

Sample output

Temperature

Top P

Top K

Maximum output tokens

Minimum output tokens

Repetition penalty

Beam count

Patches per image (JSON)

Video segments