{
  "hf_id": "nvidia/Cosmos3-Super-Text2Image",
  "meta": {
    "title": "Cosmos3-Super-Text2Image",
    "slug": "cosmos3-super-text2image",
    "provider": "NVIDIA",
    "description": "64B Cosmos3-Super specialization for high-fidelity text-to-image generation",
    "date_added": "2026-06-02",
    "date_updated": "2026-06-02",
    "tasks": [
      "omni"
    ],
    "performance_headline": "High-fidelity text-to-image — 8×H100 with CFG-parallel + ulysses + HSDP",
    "related_recipes": [
      "nvidia/Cosmos3-Super",
      "nvidia/Cosmos3-Nano",
      "nvidia/Cosmos3-Super-Image2Video"
    ],
    "hardware": {
      "h200": "verified",
      "h100": "verified",
      "gb200": "verified"
    }
  },
  "model": {
    "model_id": "nvidia/Cosmos3-Super-Text2Image",
    "min_vllm_version": "0.21.0",
    "docker_image": "vllm/vllm-omni:cosmos3",
    "architecture": "dense",
    "parameter_count": "64B",
    "active_parameters": "64B",
    "context_length": 262144,
    "base_args": [
      "--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"
    ],
    "base_env": {},
    "install": {
      "pip": {
        "command": "uv venv\nsource .venv/bin/activate\nuv pip install -U vllm --torch-backend auto"
      },
      "docker": {
        "command": "docker pull vllm/vllm-omni:cosmos3",
        "note": "Release-tested vLLM-Omni image bundling the Cosmos3 omni handlers."
      }
    }
  },
  "omni": {
    "tasks": [
      {
        "id": "t2i",
        "description": "Text → high-fidelity image"
      }
    ]
  },
  "features": {},
  "opt_in_features": [],
  "variants": {
    "default": {
      "precision": "bf16",
      "vram_minimum_gb": 154,
      "description": "BF16 weights — 8×H100 node recommended"
    }
  },
  "hardware_overrides": {},
  "guide": "## Overview\n\n**Cosmos3-Super-Text2Image** is a 64B specialization of NVIDIA's Cosmos3-Super\nomnimodal world model, tuned for **high-fidelity text-to-image generation**.\nIt is served via **vLLM-Omni**.\n\nFor best quality, text prompts should be JSON-upsampled — the model repo ships\nan `agentic_upsampling/` package and an `AGENTIC_UPSAMPLING.md` guide.\n\n## Prerequisites\n\n- 8× H100 for the documented full-node profile (4×H200 / 4×GB200 alternative below)\n- The release-tested `vllm/vllm-omni:cosmos3` container, or vLLM-Omni installed\n  on top of final `vllm==0.33.0` for this source checkout. The published\n  Cosmos3 container and its historical results use an older runtime and do\n  not qualify the current source. The separate Reasoner environment below,\n  where present, retains its own plugin/runtime requirements.\n\n## Launch command (vLLM-Omni)\n\nRecommended configuration on an 8×H100 node:\n\n```bash\ndocker pull vllm/vllm-omni:cosmos3\n\nvllm serve nvidia/Cosmos3-Super-Text2Image \\\n  --omni \\\n  --host 0.0.0.0 \\\n  --port 8000 \\\n  --cfg-parallel-size 2 \\\n  --ulysses-degree 4 \\\n  --tensor-parallel-size 1 \\\n  --use-hsdp \\\n  --hsdp-shard-size 8 \\\n  --init-timeout 1800\n```\n\nFor **4×H200 or 4×GB200**, use\n`--cfg-parallel-size 2 --ulysses-degree 2 --tensor-parallel-size 1`.\n`--enable-layerwise-offload` can reduce VRAM usage on smaller GPUs, but for\ntext-to-image it incurs a significant performance penalty.\n\n### Example: text-to-image\n\n```bash\ncurl -X POST http://localhost:8000/v1/images/generations \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\n    \"prompt\": \"a dragon flying over the Green Mountains at sunset\",\n    \"size\": \"1024x1024\",\n    \"seed\": 42\n  }' | jq -r '.data[0].b64_json' | base64 -d > output.png\n```\n\n## References\n\n- [Model card](https://huggingface.co/nvidia/Cosmos3-Super-Text2Image)\n- [Agentic prompt upsampling](https://huggingface.co/nvidia/Cosmos3-Super-Text2Image/blob/main/AGENTIC_UPSAMPLING.md)\n- [vLLM-Omni](https://github.com/vllm-project/vllm-omni)\n- [Cosmos](https://github.com/nvidia/cosmos)\n",
  "source": {
    "source": "vllm-omni",
    "repo": "vllm-project/vllm-omni",
    "ref": "main",
    "sha": "02286829737a54167e890185bcbb6948ed9afbcf",
    "path": "recipes/models/nvidia/Cosmos3-Super-Text2Image.yaml",
    "edit_url": "https://github.com/vllm-project/vllm-omni/edit/main/recipes/models/nvidia/Cosmos3-Super-Text2Image.yaml",
    "overlays": []
  }
}