MiniMax music API closed to new users: the Sume Music Router route
MiniMax stopped offering its paid Music and Lyrics APIs to new users on Aug 20, 2026. Sume's Music Router, sume/music-auto, is open: one POST, flat price.
MiniMax's music guide says that starting August 20, 2026 its paid Music Generation and Lyrics Generation APIs are no longer available to new users, and the free music APIs are discontinued. If you need a music API now, Sume's Music Router takes a prompt at POST /v1/music-router/generate, and sume/music-auto picks the engine (Lyria 3.5 today).
MiniMax facts are from its page, read 2026-09-30; Sume facts from the Music Router docs.
What exactly did MiniMax change?
Existing paying users can continue with the current API services. The free Music-3.0-free, Music-2.6-free and music-cover-free APIs are discontinued. The page points to MiniMax Audio and to the open-source MiniMax Music 3 model on Hugging Face for experimenting.
How does the Sume Music Router work?
Send a prompt of 1 to 5000 characters with an Idempotency-Key header. Omit model or set it to sume/music-auto to let Sume choose; the routable ids are sume/music-auto, lyria-3.5 and lyria-3-pro. The body is the same as Music 1.0 plus the optional model, and an unknown id fails with 400 model_not_found and a catalog_url. Read the audio from result.artifacts[] where type is audio.
| Item | Value |
|---|---|
| Endpoint | POST /v1/music-router/generate |
| Default model | sume/music-auto (Lyria 3.5 today) |
| Explicit ids | lyria-3.5, lyria-3-pro |
| Length | Steer in the prompt; duration is rejected |
| Price | Fixed Music price per audio generation, whichever model |
Is Sume a drop-in for MiniMax requests?
No. The field names differ and there is no lyrics endpoint described in these docs, so plan to rewrite the request. Put lyrics and structure in the prompt, and note that a non-empty negative_prompt is unsupported. Use GET /v1/music-router/models to list ids; see list the Music Router catalog.
What else should I check before switching?
Compare your use case against Suno alternatives and the Lyria notes in music generation API with Lyria. The engine that ran is reported in job.request.routed_model, so log it.
Sources
Related posts
More in Models
- Mirage Avatar X: a 10-second AI twin vs Sume avatar inputs
Captions' Mirage Avatar X powers its avatars and AI twins, and a twin needs ten seconds of footage. Sume avatars come from a prompt, a profile or an image.
- Nano Banana 14 reference images vs Sume's 10 input_references
Google says Gemini 3 image models mix up to 14 reference images. Sume's Nano Banana models take up to 10 input_references; ChatGPT Image 2.5 takes 16.
- Nano Banana Google Search grounding: not a Sume parameter
Google documents Search grounding for its Gemini image models. Sume lists Nano Banana 2 but no grounding field, and unlisted fields return 400.
- Nano Banana Pro aspect ratio: auto keeps the reference shape
Freepik's changelog says Nano Banana Pro accepts aspect_ratio auto to follow reference images; omitting it stays 1:1. On Sume, omitting is also not auto.
Written by Sume