DiffuCoder 7B

DiffuCoder 7B is estimated to need 5.8 GB of VRAM at Q4, 9.8 GB at Q8 and 18 GB at FP16. All figures are estimates, based on the 7.6B parameter count plus about 15% and 0.5 GB for a short context.

Yes, it fits on all four sizes. A 24 GB GPU fits at FP16 (estimated 18 GB), 16 GB and 12 GB GPUs fit at Q8 (estimated 9.8 GB) and an 8 GB GPU fits at Q4 (estimated 5.8 GB).

Yes. At Q4 the estimated VRAM is 5.8 GB. Assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is 8 GB.

Yes. DiffuCoder 7B has open weights, so you can self-host it for full control of your data.

Run DiffuCoder 7B locally

No. The input modality is text only, so images, audio, video and PDFs are not supported.

The closest alternatives, matched on model kind and release date with no comparable price, are ERNIE 4.5 0.3B by Baidu, ERNIE 4.5 21B A3B by Baidu and Jamba Large by AI21 Labs.

ERNIE 4.5 0.3BDiffuCoder 7B vs ERNIE 4.5 0.3BERNIE 4.5 21B A3BDiffuCoder 7B vs ERNIE 4.5 21B A3BJamba LargeDiffuCoder 7B vs Jamba Large

DiffuCoder 7B is open weights rather than fully open source. It is licensed under apple-amlr.

The weights are available on Hugging Face in the repository apple/DiffuCoder-7B-cpGRPO.

Run DiffuCoder 7B locally

DiffuCoder 7B was made by Apple and released on July 1, 2025.

More from Apple

Yes, DiffuCoder 7B has open weights, so it can be downloaded and self-hosted. No pricing details beyond that are given.