Skip to content
Local AI Wizard
Your machine
Theme

Cosmos3-Super-Text2Image

NVIDIA · nvidia/Cosmos3-Super-Text2Image

Weights license: OpenMDW 1.1 · code and components may use other terms

LMArena · text-to-image1,152 ± 6#3 of 14 rated here · 2026-10-07
Published repository size132.5 GBAll published files · not runtime memory
Parameters64 Bmanufacturer model card
Published components1Listed in the model card

Quality

Artificial Analysis is excluded: its free tier is internal use only.

Reported by the publisher

The publisher's card states no benchmark result for this checkpoint (read 2026-10-26). not published on the card · read 2026-10-26

Quality × memory

No comparable quality × memory pair

Image, video and audio models have no shared quality scale, and their memory comes from the publisher's documented minimum path, so quality is not charted against memory.

Consider it for

Consider Cosmos3-Super-Text2Image from NVIDIA for the text-to-image task (64B parameters).

Check first

The catalog lists the publisher's documented memory path under Runs on and has not reviewed other configurations. Actual memory depends on precision, resolution, duration and offload; consult the official source.

Runs on

Server-class only 1

No single desktop, laptop or GPU profile holds it in fast memory; memory is aggregated across GPUs.

Hardware profiles grouped by the publisher's documented minimum memory path and platform. Not a measured run.

Plan memory for this model

Measured speed

No published measurement for this checkpoint

Generation time per image is shown only when a source publishes it with hardware, build and setup. It is never estimated.

How to run

Publisher-documented memory path: Recommended 8× H100 vLLM-Omni configuration. Actual memory depends on precision, resolution, duration and offload; consult the official source. Official hardware evidence

Documented ways to run it

Copied as printed in the linked source (read 2026-10-18); not run by this site.

Setup stated first by the source · Diffusers
uv venv --python 3.13 --seed --managed-python
source .venv/bin/activate
uv pip install \
  "diffusers @ git+https://github.com/huggingface/diffusers.git" \
  accelerate \
  av \
  cosmos_guardrail \
  huggingface_hub \
  imageio \
  imageio-ffmpeg \
  torch \
  torchvision \
  transformers
Diffusers · model card · Text to image generation
import json
import torch
from diffusers import Cosmos3OmniPipeline
from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler

json_prompt = json.load(open("assets/example_caption.json"))

pipe = Cosmos3OmniPipeline.from_pretrained(
    "nvidia/Cosmos3-Super-Text2Image",
    torch_dtype=torch.bfloat16,
    device_map="cuda",
    enable_safety_checker=True,
)
pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config, flow_shift=3.0)

result = pipe(
    prompt=json.dumps(json_prompt),
    negative_prompt="",
    num_frames=1,
    height=1024,
    width=1024,
    num_inference_steps=50,
    guidance_scale=4.0,
    generator=torch.Generator(device="cuda").manual_seed(1143),
)

result.video[0].save("/tmp/cosmos3_t2i.png")
print("Saved image to /tmp/cosmos3_t2i.png")
vLLM · model card · General Invocation
vllm serve nvidia/Cosmos3-Super-Text2Image \
  --omni \
  --host 0.0.0.0 \
  --port 8000 \
  --cfg-parallel-size 2 \
  --ulysses-degree 4 \
  --tensor-parallel-size 1 \
  --use-hsdp \
  --hsdp-shard-size 8 \
  --init-timeout 1800
SGLang · model card · SGLang
git clone --branch main https://github.com/sgl-project/sglang.git
cd sglang
pip install -e "python[diffusion]"
pip install "cosmos-guardrail==0.3.1"

sglang serve \
  --model-path nvidia/Cosmos3-Super-Text2Image \
  --num-gpus 4

Evidence level: Publisher documentation linked; no execution reproduced by this site.

Before downloading
  1. Check the exact checkpoint and license linked here.
  2. Confirm a compatible runtime and precision or quantization for your platform.
  3. Allow for context, batch size and any extra media components beyond the weights.

Technical details

What the source and the catalog store for this checkpoint. Missing fields are left out, never zero.

Task and modality
text-to-image · image
Count and basis
64B · manufacturer model card
Released
nvidia/Cosmos3-Super-Text2Image model card (Hugging Face), Release Date · read 2026-10-09source
Published repository size
132.5 GBNot runtime memory or a minimum download.files
Published components
64B Mixture-of-Transformers
Runtime and precision
No local execution verified.
Commercial use
Allowed OpenMDW 1.1 text (nvidia/NV-Reason-CT LICENSE) · read 2026-10-18license text
Every stored field10 fields
Revision checked
daf3d374804be4c512c2135568a7cb95d4341d79
Architecture
Cosmos3ForConditionalGeneration
Library
cosmos
Hub task tag
text-to-image
Formats
safetensors
Repository created
2026-05-28creation is not the model's release
Repository updated
2026-09-16a commit is not a release date
Hub downloads (30 days) · likes
7,764 · 194this repository only · read 2026-10-25
Read from the Hub page
task, task tags, license name, file formatsHub overview and file listing · read 2026-10-18
Catalog state
Hub config or weights found; no inventory benchmark link

License, parameter count and release date come from the publisher's sources listed on this page (a date no source states is left empty). The hardware requirement has not been reviewed, so memory is a labelled estimate. The repository size, when shown, is the full published download. The leaderboard rows “Cosmos3-Super-Text2Image” and “Cosmos3-Super-Text2Image (agentic)” link to this one open checkpoint and share this page; no readable source maps each row to its settings.

Sources

Identity and published specifications come from the checkpoint page. Benchmark scores belong to the named release and do not measure speed on your computer.

Catalog reviewed: 2026-09-22.