zsxkib/flux-music

🎼FluxMusic Text-to-Music Generation with Rectified Flow Transformer🎶

Capabilities

Negative Prompt Seed

Community model (estimated from hardware time)

Name	Type	Description	Default	Constraints
`guidance_scale`	number	Classifier-free guidance scale	`7`	min: 0, max: 20
`model_version`	string	Select the model version to use	`"base"`	small base large giant
`negative_prompt`	string	Text prompt for negative guidance (unconditioned prompt)	`"low quality, gentle"`	—
`prompt`	string	Text prompt for music generation	`"The song is an epic blend of space-rock, rock, and post-rock genres."`	—
`save_spectrogram`	boolean	Whether to save the spectrogram image	`false`	—
`seed`	integer	Random seed. Leave blank to randomize the seed	`—`	—
`steps`	integer	Number of sampling steps	`50`	min: 1, max: 200

guidance_scale number

Classifier-free guidance scale

Default: 7 min: 0, max: 20

model_version string

Select the model version to use

Default: "base"

small base large giant

negative_prompt string

Text prompt for negative guidance (unconditioned prompt)

Default: "low quality, gentle"

prompt string

Text prompt for music generation

Default: "The song is an epic blend of space-rock, rock, and post-rock genres."

save_spectrogram boolean

Whether to save the spectrogram image

Default: false

seed integer

Random seed. Leave blank to randomize the seed

steps integer

Number of sampling steps

Default: 50 min: 1, max: 200

Version: eebfed4a1749 Updated: 2/26/2026 8.9K runs