Music Flamingo 8B

model·music understanding model·active

NVIDIA/UMD audio-language model for full-song understanding, reasoning and lyric transcription.

Recorded facts

Official site	https://huggingface.co/nvidia/music-flamingo-hf ↗
Geography	unknown
parameter count	8B
huggingface repo	nvidia/music-flamingo-hf
api	unknown
inputs	music audio; text
license	NVIDIA open-model terms in repository; human review required
outputs	text answers; captions; lyrics
weights	public
modalities	captions; lyrics; music audio; text; text answers
dependencies	unknown
model family	Music Flamingo
access method	downloadable checkpoint; Hugging Face; Gradio demo
exact version	music-flamingo-hf (8B)
repository links	https://huggingface.co/nvidia/music-flamingo-hf
evaluation results	result: Developer reports state-of-the-art aggregate performance; per-benchmark table is in paper · source: https://research.nvidia.com/labs/adlr/MF/ · benchmark: 10+ music-understanding benchmarks
evidence separation	developer claims: Authors report state-of-the-art results across more than ten music-understanding benchmarks. · measured results: · independent evaluations:
developer maintainer	NVIDIA ADLR / University of Maryland
material us relevance	Publicly accessible in or materially offered to the United States market as of snapshot.
documented capabilities	music question answering; captioning; instrument and genre identification; multilingual lyric transcription; 15-minute context
products exposing model	unknown
commercial use conditions	Commercial use not treated as verified until repository license is reviewed.
known geographic or policy restrictions	Large checkpoint and long-context compute requirements
safety watermarking disclosure provenance	unknown