Tencent · First MLX build 30 May 2026
SongGeneration
MLX
Audio other
A 5.1B parameter text to audio model from Tencent, with 10 MLX builds, from 4-bit to fp32.
- Parameters
- 5.1B From Hugging Face
- Quantisation
- Unstated SongGeneration-v2-medium
- Memory needed
- 9 GB Estimate, unified memory
- Download
- 5.6 GB Weights, estimated
- Downloads
- 99 This build, last 30 days
- Task
- Text to audio Audio
- MLX builds
- 10 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 9 GB at Unstated
8 GB
16 GB
24 GB
32 GB
64 GB
128 GB
Fits with headroomTight, close other appsWon’t fit
Estimated from the size of the weights, plus a fifth for context and working memory and 2 GB for macOS and your apps. Long contexts need more.
Run it
Terminal · Hugging Face CLI$ pip install -U huggingface_hub
$ hf download mlx-community/SongGeneration-v2-mediumMLX builds
10 on mlx-community| SongGeneration-v2-medium-4bit | 4-bit | 1.6 GB | 4 GB | 54 |
|---|---|---|---|---|
| SongGeneration-v2-large-4bit | 4-bit | 2.9 GB | 6 GB | 47 |
| SongGeneration-v2-medium-8bit | 8-bit | 3.0 GB | 6 GB | 51 |
| SongGeneration-v2-large-8bit | 8-bit | 5.4 GB | 9 GB | 58 |
| SongGeneration-v2-medium | Unstated | 5.6 GB | 9 GB | 99 |
| SongGeneration-v2-medium-bf16 | bf16 | 5.6 GB | 9 GB | 51 |
| SongGeneration-v2-large | Unstated | 10.2 GB | 14 GB | 61 |
| SongGeneration-v2-large-bf16 | bf16 | 10.2 GB | 14 GB | 49 |
| SongGeneration-v2-medium-fp32 | fp32 | 11.2 GB | 15 GB | 48 |
| SongGeneration-v2-large-fp32 | fp32 | 20.4 GB | 25 GB | 47 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.