Skip to content

Make videos on the computer you already own. FreeVideo runs MiniMax H3 in as little as 8 GB of VRAM and 16 GB of RAM, and adapts its acceleration path to your hardware.

4.8 KB
Raw Blame
{
  "vdn": {
    "url": "https://github.com/OpenVDN/vdn-minimax-h3.git",
    "commit": "30b6b380c2482f3519469350810c2955d8847fd9",
    "diffusers_subdirectory": "diffusers",
    "license": "Apache-2.0"
  },
  "diffusers": {
    "url": "https://github.com/huggingface/diffusers.git",
    "base": "3a2f35d4efa4c059c8bfb3bc0d6c906264895c81",
    "patched_tree": "37068ab7331d8b28f4cf718dba7015c742a306d2",
    "patches_from": "pinned VDN diffusers_patches/*.patch"
  },
  "encoder": {
    "comfy_url": "https://github.com/comfy-org/ComfyUI.git",
    "comfy_commit": "421a1c245c682c04d4325ba365f40c834c66f5b0",
    "implementation": "ComfyUI native H3 encoder only; no UI server or T8 custom node"
  },
  "sageattention": {
    "url": "https://github.com/thu-ml/SageAttention.git",
    "commit": "d1a57a546c3d395b1ffcbeecc66d81db76f3b4b5",
    "version": "2.2.0",
    "tested_sm120_wheel_sha256": "a89dc581440ec804408da6f6cdd84172950f79523e7dc2656fdfdc9b72a26170",
    "tested_sm120_wheel_scope": "Historical CUDA 12.9 wheel; setup builds or verifies a local wheel for the selected Torch/CUDA/GPU."
  },
  "models": {
    "vdn_repo": "OpenVDN/vdn-minimax-h3",
    "vdn_revision": "751739ee5b9e3ac802dca5d5111075fdaeb47885",
    "vdn_modelscope": {
      "repo": "OpenVDN/vdn-minimax-h3",
      "revision": "3674e3a5183da3782bdd7e57b83adb8b99a759ba",
      "hf_revision": "751739ee5b9e3ac802dca5d5111075fdaeb47885",
      "probe_file": "h3-base/transformer/diffusion_pytorch_model-00001-of-00014.safetensors"
    },
    "encoder_repo": "t8star/Vdn-Minimax-H3-Comfy",
    "encoder_revision": "d6aa5dfc669a749dc9364a36f3b6d4d1545a53e8",
    "encoder_file": "text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
    "encoder_sha256": "35a88d51044231fe332301d7a62aa81e3f2cba62febeb446e2c1e3e0ef76f2c6",
    "encoder_modelscope": {
      "repo": "LAAASSS/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
      "revision": "526349f8fbb32e5fcb0cb5e9235b30f45f1e1482",
      "hf_revision": "d6aa5dfc669a749dc9364a36f3b6d4d1545a53e8",
      "file": "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
      "probe_file": "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors"
    },
    "edge_modelscope": {
      "repo": "OpenVDN/vdn-minimax-h3-edge",
      "revision": "c247caf6f9f54fab9240e78fc7e0ec233016e47a",
      "hf_revision": "5e6ecc39f472f98ca7493000df49a4f712efb61a",
      "hf_snapshot_revision": "d95329e1441404e95082053b148ba5f8d3dd754f",
      "probe_file": "cache/blocks/00.safetensors"
    },
    "edge_modelscope_variants": {
      "rowwise": {
        "repo": "OpenVDN/vdn-minimax-h3-edge",
        "revision": "c247caf6f9f54fab9240e78fc7e0ec233016e47a",
        "hf_revision": "ae041e5aec51f8516a3416bd86d579cb29c4c796",
        "probe_file": "rowwise/cache/blocks/00.safetensors"
      },
      "int8_convrot": {
        "repo": "OpenVDN/vdn-minimax-h3-edge",
        "revision": "c247caf6f9f54fab9240e78fc7e0ec233016e47a",
        "hf_revision": "da37681f5a1f7015fdb63482142b865111b687d3",
        "probe_file": "int8-convrot/cache/blocks/00.safetensors"
      },
      "community_sigma3": {
        "repo": "OpenVDN/vdn-minimax-h3-edge",
        "hf_revision": "4bbe518a5b24aa596f1e262e5f1a08062790670b",
        "revision": "c247caf6f9f54fab9240e78fc7e0ec233016e47a"
      },
      "sampling_levels": {
        "repo": "OpenVDN/vdn-minimax-h3-edge",
        "hf_revision": "0cd66b30a58021012def78b0340caf131e0df5c2",
        "revision": "c247caf6f9f54fab9240e78fc7e0ec233016e47a"
      },
      "reference_audio": {
        "repo": "OpenVDN/vdn-minimax-h3-edge",
        "hf_revision": "b3a8dd8d4cf215b3a90bcc20b4362318f5f4f158",
        "revision": "c247caf6f9f54fab9240e78fc7e0ec233016e47a"
      },
      "sampling_steps": {
        "repo": "OpenVDN/vdn-minimax-h3-edge",
        "hf_revision": "8f6860a5e5de2844bb9af7fa801e0fafb2376bd5",
        "revision": "c247caf6f9f54fab9240e78fc7e0ec233016e47a"
      }
    },
    "latent_upscaler": {
      "repo": "LBH-123-AI/Minimax_h3_latent_Upscaler",
      "revision": "3f941d5d182014dd5c0a5e16330420ee2d4aa0c6",
      "file": "minimax_h3_latent_upscaler_3d_conv_v1/minimax_h3_latent_upscaler_3d_conv_v1_fp16.safetensors",
      "bytes": 690592672,
      "sha256": "043e5a48e161610ef6c3ea974645220354d06fa618abca15f76d084812eb55c2",
      "role": "latent_upscaler",
      "license": "MIT"
    }
  },
  "tested": {
    "python": "3.12.3",
    "torch": "2.13.0+cu130",
    "triton": "3.7.1",
    "sageattention": "2.2.0",
    "transformers": "5.15.0",
    "environment_layout": "unified",
    "scope": "SM120 kernel checks and full requests; see benchmarks/2026-09-10-unified-environment.json"
  },
  "model_license": "MiniMax H3 Community License; weights are downloaded separately",
  "tested_legacy": {
    "python": "3.12.3",
    "torch": "2.13.0+cu129",
    "triton": "3.7.1",
    "sageattention": "2.2.0",
    "flash-attn-4": "4.0.0b26"
  }
}