lucataco/videollama3-7b

VideoLLaMA 3: Frontier Multimodal Foundation Models for Video Understanding

Capabilities

Max Tokens Top-P

Community model (estimated from hardware time)

Name	Type	Description	Default	Constraints
`prompt` *	string	Text prompt to guide the model's response	`—`	—
`video` *	string (uri)	Input video file	`—`	—
`fps`	number	Frames per second to sample from video	`1`	min: 0, max: 10
`max_frames`	integer	Maximum number of frames to process	`180`	min: 0, max: 256
`max_new_tokens`	integer	Maximum number of tokens to generate	`2048`	min: 0, max: 4096
`temperature`	number	Sampling temperature	`0.2`	min: 0, max: 1
`top_p`	number	Top-p sampling	`0.9`	min: 0, max: 1

prompt required string

Text prompt to guide the model's response

video required string

Input video file

fps number

Frames per second to sample from video

Default: 1 min: 0, max: 10

max_frames integer

Maximum number of frames to process

Default: 180 min: 0, max: 256

max_new_tokens integer

Maximum number of tokens to generate

Default: 2048 min: 0, max: 4096

temperature number

Sampling temperature

Default: 0.2 min: 0, max: 1

top_p number

Top-p sampling

Default: 0.9 min: 0, max: 1

Version: 34a1f45f7068 Updated: 2/26/2026 32.7K runs