MOSS VL Realtime
MOSS-VL-Realtime is an 11B parameter vision language model built for continuous video streams. It perceives incoming frames and generates text in parallel, answers questions asked at any point in the stream, and can stay silent or revise earlier answers as new visual evidence arrives. It uses a cross attention architecture with timestamp aware frame encoding and a 256K context window.
- Type
- Multimodal
- Weights
- Open
- Downloads
- 1.3k
- Likes
- 110
- Pipeline
- video text to text
- License
- apache-2.0
Model hub
OpenMOSS-Team/MOSS-VL-Realtime
- Repo
- OpenMOSS-Team/MOSS-VL-Realtime
- Library
- transformers
- Access
- Public
Technical tags and languages
Papers
2
More from OpenMOSS
9
-
148k DL
MOSS Audio
Multimodal
MOSS-Audio is an open-source unified audio understanding model family from MOSI.AI, OpenMOSS, and the Shangha…
-
165k DL
MOSS Transcribe Diarize
Audio
MOSS Transcribe Diarize is an end to end multimodal large language model that jointly performs speech transcr…
-
86k DL
MOSS TTS Local Transformer v1.5
Audio
MOSS-TTS-Local-Transformer-v1.5 is a 5B-parameter text-to-speech model supporting 31 languages with zero-shot…
-
108k DL
MOSS TTS Nano
Audio
MOSS-TTS-Nano is an open-source multilingual tiny speech generation model from MOSI.AI and OpenMOSS. With onl…
-
145k DL
MOSS TTS v1.5
Audio
MOSS-TTS is OpenMOSS and MOSI.AI’s open-source speech and sound generation family for voice cloning, dialogue…
-
MOVA
Video
Open source foundation model that jointly generates video and audio in one pass, achieving tightly synchroniz…