Music models that run on 64 GB
4 models with published weights that fit in 64 GB — a workstation. Counted at four-bit quantisation, weights only, leaving the machine about a third of its memory and a gigabyte for context: room for roughly 67 billion parameters. A mixture of experts is counted in full, because it is held in full even though only a few experts compute.
A workstation, or a well-specified Mac. Seventy-billion-parameter models fit at four-bit, which is the class where local output stops being obviously worse than the hosted models people pay for. Loading takes real time and the machine will be warm.
Models that compose and render music from a description. They are priced per generation or per second of audio rather than per token. The question that decides whether you can use one commercially is not the price but the terms: makers differ sharply on who owns the output and whether it may be sold, and that is on the maker's page, not in the rate card.
- YuE2-3BMultimodal Art Projection3.6B≈2.4 GB at 4-bit
- ACE-Step-v1-3.5BACE-Step3.5B≈2.3 GB at 4-bit
- musicgen-melodyfacebook1.6B≈1.0 GB at 4-bit
- musicgen-smallAI at Meta0.6B≈0.4 GB at 4-bit