All tasks / Voice & audio / Music generation
Best local AI for music generation
Open-weights models and tools that run entirely on your own hardware, ranked from public benchmarks. Local tools are rated by the best open model they run.
BTop ranked: MiniMax Music 3.0 (MiniMax), 1,000 on Artificial Analysis Music Arena: Vocals (AA-Music-Vocal v1.1).7 options listed · data verified 19 Oct 2026
Ranked list
| Tier | Name | Approach | Runs | Rated by | Price | Skill |
|---|---|---|---|---|---|---|
| B |  MiniMax Music 3.0MiniMax | Models | Local | 1,000 on Artificial Analysis Music Arena: Vocals (AA-Music-Vocal v1.1) | ~24 GB VRAM | — |
| B |  Stable Audio 3 MediumStability AI | Models | Local | 996 on Artificial Analysis Music Arena: Instrumental (AA-Music-Instrumental v1.1) | Open weights | — |
| B |  ComfyUIComfy Org | Workflow platforms | Local | via MiniMax Music 3.0 | Free | advanced |
| B |  WanGP (Wan2GP)deepbeepmeep (community project) | Apps | Local | via MiniMax Music 3.0 | Free | intermediate |
| B |  audio.cpp0xShug0 (community) | Runtimes & APIs | Local | via MiniMax Music 3.0 | Free | advanced |
| C |  MusicGenMeta | Models | Local | 868 on Artificial Analysis Music Arena: Instrumental (AA-Music-Instrumental v1.1) | Open weights | — |
| C |  TTS WebUIrsxdalv (open source) | Runtimes & APIs | Local | via MusicGen | Free | intermediate |
Tiers come only from public benchmarks (Artificial Analysis Music Arena: Vocals (AA-Music-Vocal v1.1), Artificial Analysis Music Arena: Instrumental (AA-Music-Instrumental v1.1)); tools without a result of their own inherit the tier of the best model they offer. How we rank.