schema = "m64/capsule@1"
description = "SANA 1.5 1.6B text-to-image on GB10 (diffusers). SANA 1.5 1.6B at 1024x1024, 20 steps, guidance 4.5, bf16. A few seconds per image on a DGX Spark (GB10)."
kinds = ["image.generate"]
authors = ["dh7"]
license = "apache-2.0"

[hardware]
platform = "linux-aarch64"
gpu = "NVIDIA GB10"
gpu_count = 1
vram_gb = 16
driver = ">=580"
disk_gb = 10

[runtime]
sandbox = "container"
# No image: the Dockerfile builds on vllm/vllm-openai (gemma4-cu130) pinned by digest.
gpu = true
shm_gb = 8
internal_port = 9007
health = "/health"
expected_load_seconds = 180
model_name = "sana-1.5-1.6b"
user = "65534:65534"

[[artifacts]]
type = "hf"
repo = "Efficient-Large-Model/SANA1.5_1.6B_1024px_diffusers"
revision = "b77948f2b4eed5c728e9b828ccff07f7427b43cc"
include = ["*.json", "*.safetensors", "*.txt", "tokenizer*"]
mount = "/weights/sana"

[params]
size = "1024x1024"
num_inference_steps = 20
guidance_scale = 4.5


[routes]
extra = ["/v1/config"]