The SANA transformer and DC-AE VAE are Apache-2.0. The bundled gemma-2-2b-it text encoder is under the Gemma Terms of Use.
SANA 1600M 1024px — SceneWorks MLX mirror
Un-gated, repackaged mirror of Efficient-Large-Model/Sana_1600M_1024px_diffusers
for the SceneWorks native MLX worker (mlx-gen-sana).
The tensors are copied verbatim (dtype-preserving — no conversion, no
re-quantization) into the directory layout that SanaPipeline::from_snapshot
expects:
transformer/diffusion_pytorch_model.safetensors # SANA 1.6B Linear-DiT trunk (F16)
vae/diffusion_pytorch_model.safetensors # 32x DC-AE f32c32 decoder (F32)
text_encoder/gemma-2-2b-it.safetensors # gemma-2-2b-it CHI caption encoder (BF16)
text_encoder/tokenizer.json
The gemma-2-2b-it text encoder is sourced from the un-gated
SceneWorks/gemma-2-2b-it
mirror (merged from its two shards into one file) so the SANA snapshot is
self-contained.
License
The SANA transformer and DC-AE VAE weights are licensed under the
Apache License 2.0 (see LICENSE). NVIDIA relicensed them from
the earlier non-commercial NVIDIA license to Apache-2.0 on 2026-07-31 in
Efficient-Large-Model/Sana_1600M_1024px@d71cbc8e
and Efficient-Large-Model/Sana_1600M_1024px_diffusers@ac0da2ff
("Relicense model under Apache-2.0"). The weight files are byte-identical before
and after that relicence, so the grant covers the bytes mirrored here.
The bundled gemma-2-2b-it text encoder (text_encoder/ in every tier) is
provided under and subject to the Gemma Terms of Use found at
https://ai.google.dev/gemma/terms (copy: GEMMA_TERMS_OF_USE),
including the use restrictions of the Gemma Prohibited Use Policy
(https://ai.google.dev/gemma/prohibited_use_policy, copy:
GEMMA_PROHIBITED_USE_POLICY.md). Those
restrictions apply to anyone who uses or redistributes the text encoder.
See NOTICE. This mirror adds no additional grant.
Quantized