Tasks and inputs
MiniMax Music 3 is listed for Music Generation. It accepts text and produces audio. Use the publisher’s input format and preprocessing for your chosen deployment.
MiniMax Music 3 generates complete songs from a music description and optional lyrics, with a documented duration of up to five minutes.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
MiniMax Music 3 generates complete songs from a music description and optional lyrics, with a documented duration of up to five minutes.
A useful brief identifies instruments, tempo, vocal character and how the song should develop. Name the parts that matter: a restrained opening, a change in energy at the chorus or an instrumental break. The publisher's structured descriptions are intended to preserve these decisions across a longer piece.
The model combines global and local language modelling with an audio synthesis stack. The published output format is 32 kHz, 16-bit stereo WAV. That is a music-generation deliverable, rather than a conventional text-to-speech response that reads a transcript.
Use the documented serving pipeline and repository files together. The parameter count of one component does not describe the memory needs of the complete system, and a tokenizer on its own cannot generate a finished song.
Check lyric pronunciation, transitions, instrument continuity, clipping and whether the ending actually resolves. Listen on headphones and ordinary speakers. A strong opening can conceal arrangement drift or artifacts later in the track.
The publisher announced open weights on the publisher’s announced release. Confirm the repository licence and the terms of the route you use before planning distribution. No API identifier or price is invented for a local checkpoint. Compare accepted songs using the full duration and revision count, and keep this category separate from speech generation.
Publisher source 1 · Publisher source 2 Deployment limits and commercial terms belong to the selected checkpoint or API route.
Compare the model on the work you need to complete.
MiniMax Music 3 is listed for Music Generation. It accepts text and produces audio. Use the publisher’s input format and preprocessing for your chosen deployment.
Compare pronunciation, voice consistency and time to the first playable audio. Check the service’s billing unit for the actual script.
The weights are downloadable under See MiniMax Music3 repository terms. Check runtime support and size the complete model and working context for your expected concurrency.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
StepFun
StepAudio 3 Music Preview generates songs or instrumental music from a description, lyrics, vocals or reference audio.