MOSS-Image-Base + DMD2 8-step text-to-image

This service turns a text prompt into a 1024x1024 PNG. MOSS-Image-Base is OpenMOSS's base text-to-image checkpoint (Apache-2.0); this variant adds data-archetype's 8-step DMD2 LoRA, distilled on Base by distribution matching, applied unmerged through forward hooks. It uses the one sampler the LoRA was trained for (clean estimate, then re-noise with fresh noise; ComfyUI's lcm) over 8 steps at flow shift 4.4817, CFG 1 with no negative branch. It runs through diffusers with the checkpoint's own pipeline_moss_image.py (trust_remote_code) and Qwen3.5-9B as the text encoder, on one GPU, with gx10's moss_image_server.py. Generation settings are fixed by this capsule (size "1024x1024", num_inference_steps 8, guidance_scale 1.0, shift 4.4817, sampler "fresh-noise", lora "moss_image_base_dmd2"). You send a prompt and, optionally, a seed.

Base URL: {{BASE_URL}}

Routes

POST /v1/images/generations

OpenAI-style image generation. Returns one image as base64 PNG.

curl -s {{BASE_URL}}/v1/images/generations {{AUTH}} \
  -H "Content-Type: application/json" \
  -d '{"prompt": "a lighthouse at sunset", "seed": 42}' \
  | python3 -c 'import sys,json,base64; d=json.load(sys.stdin); open("out.png","wb").write(base64.b64decode(d["data"][0]["b64_json"])); print(d["params_used"])'

Request fields:

• prompt (string, required): what to draw.
• seed (integer >= 0, optional): the random seed. The same prompt and seed give the same image. Leave it out for a random one.

Also accepted, for stock OpenAI clients: model (any string), n (must be 1), response_format (must be "b64_json"), user, and size if it is "1024x1024". Any other field, or another value for a fixed one, is rejected with 400.

The response is {"created": ..., "data": [{"b64_json": "<PNG, base64>"}], "params_used": {...}}.

The OpenAI Python SDK works too:

from openai import OpenAI
client = OpenAI(base_url="{{BASE_URL}}/v1", api_key="unused")
r = client.images.generate(model="moss-image-base-dmd2", prompt="a lighthouse at sunset",
                           response_format="b64_json", extra_body={"seed": 7})

GET /v1/config

The server's settings and what they mean.

GET /status

Where the service is in its start-up: phase is one of building, downloading, starting, loading, ready, error, with eta_seconds.

Timings and concurrency

• First start: builds a container image and downloads about 31 GB of weights. After that, loading takes about LOADTIME (budgeted at 600 s).
• Each image takes about IMGTIME on a DGX Spark (GB10).
• Send requests one at a time, and allow at least IMGTIMEOUT per request in your client timeout.

Errors

• 503 with Retry-After while the service is still starting.
• 400: a field was rejected; the body's error says which and why.
• 422: prompt missing or a field of the wrong type.
• 5xx otherwise: retry once, then check GET /status.