{
 "generated": "2026-11-14 (America/Chicago)",
 "method": {
  "memory": "Estimated memory = resident weight bytes x 1.05 + 4 GB (runtime + about 32K tokens of KV cache) for LLMs; media = files + 4 GB, an upper bound since ComfyUI/Diffusers offload components. Estimates, not measurements.",
  "tok_s": "Token/s = memory bandwidth / (active params x bits-per-weight / 8). Bits-per-weight = GGUF bytes x 8 / GGUF parameter count. This is a theoretical ceiling for generation; real numbers are usually well below it. Two-box figures take off 18% for RPC only when the model does not fit one box (otherwise run it on one box). Not given where active params are unknown or the model needs conversion.",
  "verdicts": "fits = fits in the GPU-usable budget; offload = needs raised limits or CPU/RAM offload; convert = original safetensors, quantize first; after-RAM-upgrade = MoE expert offload once omarchy has 128 GB; unknown = machine specs unknown."
 },
 "fleet": [
  {
   "id": "m5x1",
   "name": "Bosgame M5 (x1)",
   "status": "planned",
   "cpu": "AMD Ryzen AI Max+ 395 (16C/32T)",
   "gpu": "Radeon 8060S iGPU (40 CU, RDNA 3.5)",
   "ram": "128 GB LPDDR5X unified",
   "bw_gbs": 215,
   "bw_note": "256 GB/s on paper, ~210-220 measured",
   "budget_gb": 108,
   "offload_gb": 118,
   "budget_note": "~108-110 GB GPU-usable on Linux with GTT raised (96 GB cap on Windows). Fit budget used here: 108 GB.",
   "src": "MANIFEST hardware notes; AI Budget Build more-hardware.json"
  },
  {
   "id": "m5x2",
   "name": "Bosgame M5 (x2, RPC)",
   "status": "planned",
   "cpu": "2x Ryzen AI Max+ 395",
   "gpu": "2x Radeon 8060S over llama.cpp RPC (USB4 or 10GbE)",
   "ram": "2x128 GB",
   "bw_gbs": 215,
   "bw_note": "Per box; linking adds capacity, not speed (community: ~14-22% token-gen cost)",
   "budget_gb": 210,
   "offload_gb": 225,
   "budget_note": "~200-225 GB usable across two nodes. Fit budget used here: 210 GB; 210-225 GB = tight.",
   "src": "MANIFEST hardware notes (aifinitee, WillItRunAI)"
  },
  {
   "id": "mbp",
   "name": "MacBook Pro (M4 Max)",
   "status": "owned",
   "cpu": "Apple M4 Max, 14 cores (10P+4E)",
   "gpu": "32-core GPU, Metal 4",
   "ram": "36 GB unified",
   "bw_gbs": 410,
   "bw_note": "Apple spec for 14-core M4 Max",
   "budget_gb": 27,
   "offload_gb": 32,
   "budget_note": "macOS caps GPU wired memory by default (~75% = ~27 GB). Raising iogpu.wired_limit_mb to ~32 GB is possible but leaves little for the OS.",
   "src": "system_profiler on the machine (Nov 14 2026)"
  },
  {
   "id": "mini",
   "name": "Mac mini (M4)",
   "status": "owned",
   "cpu": "Apple M4",
   "gpu": "Apple M4 GPU (core count unknown)",
   "ram": "16 GB unified",
   "bw_gbs": 120,
   "bw_note": "Apple spec for M4",
   "budget_gb": 10.5,
   "offload_gb": 12,
   "budget_note": "~2/3 of 16 GB usable by the GPU by default. Already busy as build box + Hermes/Ollama.",
   "src": "sysctl on the machine (Nov 14 2026)"
  },
  {
   "id": "omarchy",
   "name": "omarchy (AMD desktop)",
   "status": "owned",
   "cpu": "AMD Ryzen 9 5950X",
   "gpu": "Radeon RX 6600 XT 8 GB (RDNA2, gfx1032)",
   "ram": "Current RAM unknown (128 GB DDR4 upgrade planned)",
   "bw_gbs": 256,
   "bw_note": "GPU VRAM 256 GB/s; dual-channel DDR4 system RAM is far slower (~50 GB/s class, estimate)",
   "budget_gb": 7,
   "offload_gb": null,
   "budget_note": "7 GB of the 8 GB VRAM for weights. Offload capacity depends on system RAM, which is unknown today; after the 128 GB upgrade ~110 GB of MoE experts could sit in RAM.",
   "src": "AI Budget Build my-pcs.json + gpus.json"
  },
  {
   "id": "nas",
   "name": "Ugreen DXP2800 NAS",
   "status": "owned",
   "cpu": "Intel N100 (4 cores)",
   "gpu": "Intel UHD (iGPU)",
   "ram": "8 GB",
   "bw_gbs": null,
   "bw_note": "Single-channel DDR5; not a model runner",
   "budget_gb": 3,
   "offload_gb": null,
   "budget_note": "Storage host. Already runs ollama/open-webui containers; only tiny models (e.g. Whisper on CPU) are realistic.",
   "src": "/proc/cpuinfo + free on the NAS"
  },
  {
   "id": "neo",
   "name": "MacBook Neo",
   "status": "owned",
   "cpu": "unknown",
   "gpu": "unknown",
   "ram": "unknown",
   "bw_gbs": null,
   "bw_note": "unknown",
   "budget_gb": null,
   "offload_gb": null,
   "budget_note": "Specs unknown: machine was offline/unreachable on Nov 14 2026.",
   "src": "unreachable"
  }
 ],
 "models": [
  {
   "repo": "Qwen/Qwen3.8-27B",
   "hf_url": "https://huggingface.co/Qwen/Qwen3.8-27B",
   "publisher": "Qwen",
   "family": "Qwen3.8-27B",
   "modality": "text+vision LLM",
   "kind": "official",
   "params_total_b": 27.0,
   "params_active_b": 27.0,
   "moe": false,
   "params_note": "27B dense (model card). Hybrid attention: 48 linear + 16 full layers (config.json). Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-27B",
   "quant_format": "BF16 safetensors",
   "size_bytes": 55586107015,
   "size_gib": 51.77,
   "resident_bytes": 55586107015,
   "license": "apache-2.0",
   "abliterated": false,
   "public_folder": "official",
   "status": "done",
   "engine": "Transformers / vLLM / SGLang; convert for llama.cpp",
   "notes": "Original weights. Archive copy.",
   "context_native": 262144,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 62.4,
   "id": "qwen3-8-27b-bf16"
  },
  {
   "repo": "Qwen/Qwen3.8-Flash-Next",
   "hf_url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
   "publisher": "Qwen",
   "family": "Qwen3.8 Flash-Next",
   "modality": "text+vision LLM",
   "kind": "official",
   "params_total_b": 125.0,
   "params_active_b": 6.0,
   "moe": true,
   "params_note": "125B MoE with 6B activated, plus 51B n-gram embedding and 4B MTP (model card). 512 experts, 10 routed + 1 shared active. Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
   "quant_format": "BF16 safetensors",
   "size_bytes": 360023343554,
   "size_gib": 335.3,
   "resident_bytes": 360023343554,
   "license": "qwen-community-1.0",
   "abliterated": false,
   "public_folder": "official",
   "status": "done",
   "engine": "Transformers / vLLM / SGLang",
   "notes": "Original weights. Archive copy.",
   "context_native": 262144,
   "license_note": "Qwen Community License 1.0: commercial OK; MaaS/AI-assistant businesses need a separate license; attribution above 100M MAU or $20M/month.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 382.0,
   "id": "qwen3-8-flash-next-bf16"
  },
  {
   "repo": "zai-org/GLM-5.3-Flash",
   "hf_url": "https://huggingface.co/zai-org/GLM-5.3-Flash",
   "publisher": "zai-org",
   "family": "GLM-5.3-Flash",
   "modality": "text+vision LLM",
   "kind": "official",
   "params_total_b": 320.0,
   "params_active_b": 18.0,
   "moe": true,
   "params_note": "320B total, 18B active (model card). 288 routed experts, 8 per token; 34 linear + 11 sparse-attention layers (config.json).",
   "params_source": "https://huggingface.co/zai-org/GLM-5.3-Flash",
   "quant_format": "safetensors (as published)",
   "size_bytes": 328366173639,
   "size_gib": 305.81,
   "resident_bytes": 328366173639,
   "license": "mit",
   "abliterated": false,
   "public_folder": "official",
   "status": "done",
   "engine": "Transformers / vLLM / SGLang",
   "notes": "Original weights. Archive copy.",
   "context_native": 1048576,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 348.8,
   "id": "glm-5-3-flash-safetensors"
  },
  {
   "repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
   "hf_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
   "publisher": "deepseek-ai",
   "family": "DeepSeek V4.1 Flash 1017",
   "modality": "text+vision LLM",
   "kind": "official",
   "params_total_b": 763.2,
   "params_active_b": null,
   "moe": true,
   "params_note": "763.2B per HF safetensors count, including 196.6B of Engram lookup tables (ISTA-DASLab card). 384 routed experts, 6 per token (config.json). Active parameter count is not stated on the official card: unknown.",
   "params_source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
   "quant_format": "FP8/FP4 safetensors (ISTA card calls base FP8/FP4)",
   "size_bytes": 510311062017,
   "size_gib": 475.26,
   "resident_bytes": 510311062017,
   "license": "mit",
   "abliterated": false,
   "public_folder": "official",
   "status": "done",
   "engine": "vLLM / SGLang",
   "notes": "Original weights. Archive copy.",
   "context_native": 1048576,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "convert",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 539.8,
   "id": "deepseek-v4-1-flash-1017-fp8-fp4"
  },
  {
   "repo": "unsloth/Qwen3.8-27B-GGUF",
   "hf_url": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF",
   "publisher": "unsloth",
   "family": "Qwen3.8-27B",
   "modality": "text+vision LLM",
   "kind": "gguf",
   "params_total_b": 27.0,
   "params_active_b": 27.0,
   "moe": false,
   "params_note": "27B dense (model card). Hybrid attention: 48 linear + 16 full layers (config.json). Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-27B",
   "quant_format": "UD-Q8_K_XL GGUF + mmproj",
   "size_bytes": 32389149295,
   "size_gib": 30.16,
   "resident_bytes": 32389149295,
   "license": "apache-2.0",
   "abliterated": false,
   "public_folder": "official",
   "status": "done",
   "engine": "llama.cpp / LM Studio / Ollama",
   "notes": "Near-lossless 8-bit.",
   "context_native": 262144,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 27.32,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": 6.7
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 6.7
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 38.0,
   "id": "qwen3-8-27b-gguf-ud-q8-k-xl"
  },
  {
   "repo": "unsloth/Qwen3.8-Flash-Next-GGUF",
   "hf_url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF",
   "publisher": "unsloth",
   "family": "Qwen3.8 Flash-Next",
   "modality": "text+vision LLM",
   "kind": "gguf",
   "params_total_b": 125.0,
   "params_active_b": 6.0,
   "moe": true,
   "params_note": "125B MoE with 6B activated, plus 51B n-gram embedding and 4B MTP (model card). 512 experts, 10 routed + 1 shared active. Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
   "quant_format": "UD-IQ4_XS GGUF (4 shards) + mmproj",
   "size_bytes": 94579972258,
   "size_gib": 88.08,
   "resident_bytes": 94579972258,
   "license": "qwen-community-1.0",
   "abliterated": false,
   "public_folder": "official",
   "status": "in-progress",
   "engine": "llama.cpp / LM Studio",
   "notes": "One-box pick for Flash-Next.",
   "context_native": 262144,
   "license_note": "Qwen Community License 1.0: commercial OK; MaaS/AI-assistant businesses need a separate license; attribution above 100M MAU or $20M/month.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 176.94,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": 67.0
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 67.0
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 103.3,
   "id": "qwen3-8-flash-next-gguf-ud-iq4-xs"
  },
  {
   "repo": "unsloth/GLM-5.3-Flash-GGUF",
   "hf_url": "https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF",
   "publisher": "unsloth",
   "family": "GLM-5.3-Flash",
   "modality": "text+vision LLM",
   "kind": "gguf",
   "params_total_b": 320.0,
   "params_active_b": 18.0,
   "moe": true,
   "params_note": "320B total, 18B active (model card). 288 routed experts, 8 per token; 34 linear + 11 sparse-attention layers (config.json).",
   "params_source": "https://huggingface.co/zai-org/GLM-5.3-Flash",
   "quant_format": "UD-IQ2_XXS GGUF + mmproj",
   "size_bytes": 103008979940,
   "size_gib": 95.93,
   "resident_bytes": 103008979940,
   "license": "mit",
   "abliterated": false,
   "public_folder": "official",
   "status": "queued",
   "engine": "llama.cpp",
   "notes": "Tight one-box pick (2-bit).",
   "context_native": 1048576,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 320.76,
   "fit": {
    "m5x1": {
     "verdict": "offload",
     "est_tok_s": 37.2
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 30.5
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 112.2,
   "id": "glm-5-3-flash-gguf-ud-iq2-xxs"
  },
  {
   "repo": "ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF",
   "hf_url": "https://huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF",
   "publisher": "ISTA-DASLab",
   "family": "DeepSeek V4.1 Flash 1017",
   "modality": "text+vision LLM",
   "kind": "gguf",
   "params_total_b": 763.2,
   "params_active_b": null,
   "moe": true,
   "params_note": "763.2B per HF safetensors count, including 196.6B of Engram lookup tables (ISTA-DASLab card). 384 routed experts, 6 per token (config.json). Active parameter count is not stated on the official card: unknown.",
   "params_source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
   "quant_format": "GSQ-RCO IQ1_M backbone + shared Engram IQ4_NL + mmproj",
   "size_bytes": 232288767636,
   "size_gib": 216.34,
   "resident_bytes": 120720000000.0,
   "license": "mit",
   "abliterated": false,
   "public_folder": "official",
   "status": "queued",
   "engine": "llama.cpp (build from Oct 27 2026 or later)",
   "notes": "Backbone 120.72 GB must be resident; the 110.60 GB Engram file can stay memory-mapped on SSD (ISTA card). Engram must be linked as shard 6.",
   "context_native": 1048576,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "no",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 130.8,
   "id": "deepseek-v4-1-flash-1017-gsq-rco-gguf-gsq-rco"
  },
  {
   "repo": "alesha-pro/Qwen3.8-27B-S-mirai-GGUF",
   "hf_url": "https://huggingface.co/alesha-pro/Qwen3.8-27B-S-mirai-GGUF",
   "publisher": "alesha-pro",
   "family": "Qwen3.8-27B",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 27.0,
   "params_active_b": 27.0,
   "moe": false,
   "params_note": "27B dense (model card). Hybrid attention: 48 linear + 16 full layers (config.json). Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-27B",
   "quant_format": "Mirai S 2.4-bit GGUF + mmproj + refusal control vector",
   "size_bytes": 12102272287,
   "size_gib": 11.27,
   "resident_bytes": 12102272287,
   "license": "apache-2.0",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "done",
   "engine": "mirai-s-ada or llama.cpp-mirai-s fork ONLY",
   "notes": "Card: mainline llama.cpp, LM Studio and Ollama will not load this file. Abliteration is applied via a refusal-direction control vector.",
   "context_native": 262144,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 16.7,
   "id": "qwen3-8-27b-s-mirai-gguf-mirai"
  },
  {
   "repo": "alesha-pro/Qwen3.8-Flash-Next-abliterated-GSQ-RCO-Strata-GGUF",
   "hf_url": "https://huggingface.co/alesha-pro/Qwen3.8-Flash-Next-abliterated-GSQ-RCO-Strata-GGUF",
   "publisher": "alesha-pro",
   "family": "Qwen3.8 Flash-Next",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 125.0,
   "params_active_b": 6.0,
   "moe": true,
   "params_note": "125B MoE with 6B activated, plus 51B n-gram embedding and 4B MTP (model card). 512 experts, 10 routed + 1 shared active. Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
   "quant_format": "GSQ-RCO IQ3_S GGUF + mmproj + refusal vector",
   "size_bytes": 84525749351,
   "size_gib": 78.72,
   "resident_bytes": 84525749351,
   "license": "qwen-community-license-1.0",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "done",
   "engine": "Strata engine (setup per card)",
   "notes": "Card reports peak 23.8 GB VRAM on its test rig with Strata offload.",
   "context_native": 262144,
   "license_note": "Qwen Community License 1.0 (inherited from Flash-Next).",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 176.94,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": 75.0
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 75.0
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 92.8,
   "id": "qwen3-8-flash-next-abliterated-gsq-rco-strata-gguf-gsq-rco"
  },
  {
   "repo": "alesha-pro/GLM-5.3-Flash-abliterated-Maya-M-GGUF",
   "hf_url": "https://huggingface.co/alesha-pro/GLM-5.3-Flash-abliterated-Maya-M-GGUF",
   "publisher": "alesha-pro",
   "family": "GLM-5.3-Flash",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 320.0,
   "params_active_b": 18.0,
   "moe": true,
   "params_note": "320B total, 18B active (model card). 288 routed experts, 8 per token; 34 linear + 11 sparse-attention layers (config.json).",
   "params_source": "https://huggingface.co/zai-org/GLM-5.3-Flash",
   "quant_format": "Maya-M IQ2_S GGUF + vision + refusal vector",
   "size_bytes": 117143259589,
   "size_gib": 109.1,
   "resident_bytes": 117143259589,
   "license": "mit",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "done",
   "engine": "Project Maya engine + alesha-pro patch ONLY",
   "notes": "Not mainline llama.cpp. Card: 56 tok/s on 4x RTX 3090 (author's rig).",
   "context_native": 1048576,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 320.76,
   "fit": {
    "m5x1": {
     "verdict": "no",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 127.0,
   "id": "glm-5-3-flash-abliterated-maya-m-gguf-maya-m"
  },
  {
   "repo": "huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF",
   "hf_url": "https://huggingface.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF",
   "publisher": "huihui-ai",
   "family": "Qwen3.8-27B",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 27.0,
   "params_active_b": 27.0,
   "moe": false,
   "params_note": "27B dense (model card). Hybrid attention: 48 linear + 16 full layers (config.json). Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-27B",
   "quant_format": "Q4_K GGUF + mmproj",
   "size_bytes": 17741875547,
   "size_gib": 16.52,
   "resident_bytes": 17741875547,
   "license": "apache-2.0",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "done",
   "engine": "llama.cpp / LM Studio / Ollama",
   "notes": "Small abliterated 27B.",
   "context_native": 262144,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 27.32,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": 12.3
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 12.3
    },
    "mbp": {
     "verdict": "fits",
     "est_tok_s": 23.4
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 22.6,
   "id": "huihui-qwen3-8-27b-abliterated-gguf-q4-k"
  },
  {
   "repo": "huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF",
   "hf_url": "https://huggingface.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF",
   "publisher": "huihui-ai",
   "family": "Qwen3.8-27B",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 27.0,
   "params_active_b": 27.0,
   "moe": false,
   "params_note": "27B dense (model card). Hybrid attention: 48 linear + 16 full layers (config.json). Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-27B",
   "quant_format": "BF16 GGUF + mmproj",
   "size_bytes": 55588895067,
   "size_gib": 51.77,
   "resident_bytes": 55588895067,
   "license": "apache-2.0",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "done",
   "engine": "llama.cpp",
   "notes": "Full-precision abliterated 27B (source for your own quants).",
   "context_native": 262144,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 27.32,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": 3.9
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 3.9
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 62.4,
   "id": "huihui-qwen3-8-27b-abliterated-gguf-bf16"
  },
  {
   "repo": "huihui-ai/Huihui-Qwen3.8-Flash-Next-abliterated-v2-GGUF",
   "hf_url": "https://huggingface.co/huihui-ai/Huihui-Qwen3.8-Flash-Next-abliterated-v2-GGUF",
   "publisher": "huihui-ai",
   "family": "Qwen3.8 Flash-Next",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 125.0,
   "params_active_b": 6.0,
   "moe": true,
   "params_note": "125B MoE with 6B activated, plus 51B n-gram embedding and 4B MTP (model card). 512 experts, 10 routed + 1 shared active. Native context 262,144.",
   "params_source": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
   "quant_format": "UD-Q4_K_XL V2 GGUF (4 shards) + mmproj",
   "size_bytes": 113424100802,
   "size_gib": 105.63,
   "resident_bytes": 113424100802,
   "license": "qwen-community-1.0",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "done",
   "engine": "llama.cpp",
   "notes": "Slightly over one-box comfort; fine on two boxes.",
   "context_native": 262144,
   "license_note": "Qwen Community License 1.0: commercial OK; MaaS/AI-assistant businesses need a separate license; attribution above 100M MAU or $20M/month.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 176.94,
   "fit": {
    "m5x1": {
     "verdict": "no",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 45.8
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 123.1,
   "id": "huihui-qwen3-8-flash-next-abliterated-v2-gguf-ud-q4-k-xl"
  },
  {
   "repo": "huihui-ai/Huihui-GLM-5.3-Flash-abliterated-GGUF",
   "hf_url": "https://huggingface.co/huihui-ai/Huihui-GLM-5.3-Flash-abliterated-GGUF",
   "publisher": "huihui-ai",
   "family": "GLM-5.3-Flash",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 320.0,
   "params_active_b": 18.0,
   "moe": true,
   "params_note": "320B total, 18B active (model card). 288 routed experts, 8 per token; 34 linear + 11 sparse-attention layers (config.json).",
   "params_source": "https://huggingface.co/zai-org/GLM-5.3-Flash",
   "quant_format": "UD-IQ2_XXS GGUF + mmproj",
   "size_bytes": 103008971277,
   "size_gib": 95.93,
   "resident_bytes": 103008971277,
   "license": "mit",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "queued",
   "engine": "llama.cpp",
   "notes": "Abliterated twin of the unsloth one-box GLM pick.",
   "context_native": 1048576,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": 320.76,
   "fit": {
    "m5x1": {
     "verdict": "offload",
     "est_tok_s": 37.2
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": 30.5
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "after-RAM-upgrade",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 112.2,
   "id": "huihui-glm-5-3-flash-abliterated-gguf-ud-iq2-xxs"
  },
  {
   "repo": "huihui-ai/Huihui-DeepSeek-V4.1-Flash-1017-abliterated-GGUF",
   "hf_url": "https://huggingface.co/huihui-ai/Huihui-DeepSeek-V4.1-Flash-1017-abliterated-GGUF",
   "publisher": "huihui-ai",
   "family": "DeepSeek V4.1 Flash 1017",
   "modality": "text+vision LLM",
   "kind": "abliterated",
   "params_total_b": 763.2,
   "params_active_b": null,
   "moe": true,
   "params_note": "763.2B per HF safetensors count, including 196.6B of Engram lookup tables (ISTA-DASLab card). 384 routed experts, 6 per token (config.json). Active parameter count is not stated on the official card: unknown.",
   "params_source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
   "quant_format": "UD-IQ1_M GGUF (5 shards) + mmproj",
   "size_bytes": 251376566398,
   "size_gib": 234.11,
   "resident_bytes": 251376566398,
   "license": "mit",
   "abliterated": true,
   "public_folder": "abliterated",
   "status": "done",
   "engine": "llama.cpp (huihui build notes on card)",
   "notes": "Card: only 18 attn_output tensors ablated. Residency of Engram tensors under mmap not verified, so treated as fully resident.",
   "context_native": 1048576,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "no",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 267.9,
   "id": "huihui-deepseek-v4-1-flash-1017-abliterated-gguf-ud-iq1-m"
  },
  {
   "repo": "openai/whisper-large-v3-turbo",
   "hf_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
   "publisher": "openai",
   "family": "Whisper large-v3-turbo",
   "modality": "speech-to-text",
   "kind": "media",
   "params_total_b": 0.809,
   "params_active_b": null,
   "moe": false,
   "params_note": "809M params (HF safetensors count).",
   "params_source": "https://huggingface.co/openai/whisper-large-v3-turbo",
   "quant_format": "safetensors",
   "size_bytes": 1622466054,
   "size_gib": 1.51,
   "resident_bytes": 1622466054,
   "license": "mit",
   "abliterated": false,
   "public_folder": "media",
   "status": "queued",
   "engine": "Transformers / openai-whisper / whisper.cpp (own ggml file)",
   "notes": "809M params (HF safetensors count).",
   "context_native": null,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": "On the NAS this is CPU-only and slow (Intel N100).",
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "offload",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 5.6,
   "id": "whisper-large-v3-turbo-safetensors"
  },
  {
   "repo": "Qwen/Qwen3-TTS-12Hz-1.7B-Base",
   "hf_url": "https://huggingface.co/Qwen/Qwen3-TTS-12Hz-1.7B-Base",
   "publisher": "Qwen",
   "family": "Qwen3-TTS 1.7B Base",
   "modality": "text-to-speech (voice clone)",
   "kind": "media",
   "params_total_b": 1.93,
   "params_active_b": null,
   "moe": false,
   "params_note": "1.93B per HF safetensors count; card: 3-second voice clone, 10 languages.",
   "params_source": "https://huggingface.co/Qwen/Qwen3-TTS-12Hz-1.7B-Base",
   "quant_format": "safetensors",
   "size_bytes": 4544229700,
   "size_gib": 4.23,
   "resident_bytes": 4544229700,
   "license": "apache-2.0",
   "abliterated": false,
   "public_folder": "media",
   "status": "queued",
   "engine": "qwen-tts (pip)",
   "notes": "1.93B per HF safetensors count; card: 3-second voice clone, 10 languages.",
   "context_native": null,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 8.5,
   "id": "qwen3-tts-12hz-1-7b-base-safetensors"
  },
  {
   "repo": "fishaudio/s2-pro",
   "hf_url": "https://huggingface.co/fishaudio/s2-pro",
   "publisher": "fishaudio",
   "family": "Fish Audio S2 Pro",
   "modality": "text-to-speech",
   "kind": "media",
   "params_total_b": 4.56,
   "params_active_b": null,
   "moe": false,
   "params_note": "Card: 4B slow AR + 400M fast AR. HF safetensors count 4.56B.",
   "params_source": "https://huggingface.co/fishaudio/s2-pro",
   "quant_format": "safetensors",
   "size_bytes": 11011629649,
   "size_gib": 10.26,
   "resident_bytes": 11011629649,
   "license": "fish-audio-research-license",
   "abliterated": false,
   "public_folder": "media",
   "status": "queued",
   "engine": "fish-speech",
   "notes": "Card: 4B slow AR + 400M fast AR. HF safetensors count 4.56B.",
   "context_native": null,
   "license_note": "Research/non-commercial free; commercial use needs a separate Fish Audio license (card).",
   "non_commercial": true,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 15.0,
   "id": "s2-pro-safetensors"
  },
  {
   "repo": "ACE-Step/Ace-Step1.5",
   "hf_url": "https://huggingface.co/ACE-Step/Ace-Step1.5",
   "publisher": "ACE-Step",
   "family": "ACE-Step 1.5",
   "modality": "text-to-music",
   "kind": "media",
   "params_total_b": null,
   "params_active_b": null,
   "moe": false,
   "params_note": "Card: runs locally with under 4 GB VRAM. Total params not stated on card: unknown.",
   "params_source": "https://huggingface.co/ACE-Step/Ace-Step1.5",
   "quant_format": "safetensors (turbo DiT + LM)",
   "size_bytes": 10092102593,
   "size_gib": 9.4,
   "resident_bytes": 10092102593,
   "license": "mit",
   "abliterated": false,
   "public_folder": "media",
   "status": "queued",
   "engine": "ACE-Step-1.5 (GitHub)",
   "notes": "Card: runs locally with under 4 GB VRAM. Total params not stated on card: unknown.",
   "context_native": null,
   "license_note": "MIT: commercial OK.",
   "non_commercial": false,
   "fit_note": "Card: runs locally with less than 4 GB VRAM.",
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 8.0,
   "id": "ace-step1-5-safetensors"
  },
  {
   "repo": "Comfy-Org/YuE2",
   "hf_url": "https://huggingface.co/Comfy-Org/YuE2",
   "publisher": "Comfy-Org",
   "family": "YuE2 (ComfyUI repack)",
   "modality": "lyrics-to-music",
   "kind": "media",
   "params_total_b": 3.0,
   "params_active_b": null,
   "moe": false,
   "params_note": "Base model m-a-p/YuE2-3B (card metadata).",
   "params_source": "https://huggingface.co/Comfy-Org/YuE2",
   "quant_format": "bf16 safetensors (ComfyUI)",
   "size_bytes": 13147793678,
   "size_gib": 12.24,
   "resident_bytes": 13147793678,
   "license": "cc-by-nc-4.0",
   "abliterated": false,
   "public_folder": "media",
   "status": "queued",
   "engine": "ComfyUI",
   "notes": "Base model m-a-p/YuE2-3B (card metadata).",
   "context_native": null,
   "license_note": "Non-commercial only (CC BY-NC 4.0).",
   "non_commercial": true,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 17.1,
   "id": "yue2-bf16"
  },
  {
   "repo": "Qwen/Qwen-Image-2.1",
   "hf_url": "https://huggingface.co/Qwen/Qwen-Image-2.1",
   "publisher": "Qwen",
   "family": "Qwen-Image 2.1",
   "modality": "text-to-image + editing",
   "kind": "media",
   "params_total_b": 7.1,
   "params_active_b": null,
   "moe": false,
   "params_note": "Card: 7B visual generation component (32 single-stream DiT layers); repo also holds the text encoder.",
   "params_source": "https://huggingface.co/Qwen/Qwen-Image-2.1",
   "quant_format": "safetensors (as published)",
   "size_bytes": 33134949561,
   "size_gib": 30.86,
   "resident_bytes": 33134949561,
   "license": "qwen-research",
   "abliterated": false,
   "public_folder": "usb-media/image",
   "status": "in-progress",
   "engine": "Diffusers (QwenImage21Pipeline)",
   "notes": "Card: 7B visual generation component (32 single-stream DiT layers); repo also holds the text encoder.",
   "context_native": null,
   "license_note": "Qwen Research License: NON-COMMERCIAL purposes only (LICENSE file).",
   "non_commercial": true,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 37.1,
   "id": "qwen-image-2-1-safetensors"
  },
  {
   "repo": "OpenMOSS-Team/MOSS-Image-Base",
   "hf_url": "https://huggingface.co/OpenMOSS-Team/MOSS-Image-Base",
   "publisher": "OpenMOSS-Team",
   "family": "MOSS-Image Base",
   "modality": "text-to-image",
   "kind": "media",
   "params_total_b": null,
   "params_active_b": null,
   "moe": false,
   "params_note": "Param count not stated on card: unknown (HF safetensors count 5.08B for part of the repo).",
   "params_source": "https://huggingface.co/OpenMOSS-Team/MOSS-Image-Base",
   "quant_format": "safetensors (as published)",
   "size_bytes": 27229060663,
   "size_gib": 25.36,
   "resident_bytes": 27229060663,
   "license": "apache-2.0",
   "abliterated": false,
   "public_folder": "usb-media/image",
   "status": "queued",
   "engine": "Diffusers / ComfyUI (MOSS-Image plugin)",
   "notes": "Param count not stated on card: unknown (HF safetensors count 5.08B for part of the repo).",
   "context_native": null,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "offload",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 31.2,
   "id": "moss-image-base-safetensors"
  },
  {
   "repo": "Wan-AI/Wan2.2-T2V-A14B",
   "hf_url": "https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B",
   "publisher": "Wan-AI",
   "family": "Wan 2.2 T2V A14B",
   "modality": "text-to-video",
   "kind": "media",
   "params_total_b": null,
   "params_active_b": null,
   "moe": false,
   "params_note": "Card: MoE video diffusion (high-noise + low-noise experts, A14B), 5 s clips at 480P/720P. Total params not stated: unknown.",
   "params_source": "https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B",
   "quant_format": "safetensors (as published)",
   "size_bytes": 126201624837,
   "size_gib": 117.53,
   "resident_bytes": 126201624837,
   "license": "apache-2.0",
   "abliterated": false,
   "public_folder": "usb-media/video",
   "status": "queued",
   "engine": "Wan2.2 generate.py / ComfyUI",
   "notes": "Card: MoE video diffusion (high-noise + low-noise experts, A14B), 5 s clips at 480P/720P. Total params not stated: unknown.",
   "context_native": null,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": "Card's single-GPU command uses --offload_model True --convert_model_dtype; high-noise and low-noise experts are used one after the other, so peak memory is well below the repo size.",
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "offload",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "offload",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 130.2,
   "id": "wan2-2-t2v-a14b-safetensors"
  },
  {
   "repo": "Wan-AI/Wan2.2-I2V-A14B",
   "hf_url": "https://huggingface.co/Wan-AI/Wan2.2-I2V-A14B",
   "publisher": "Wan-AI",
   "family": "Wan 2.2 I2V A14B",
   "modality": "image-to-video",
   "kind": "media",
   "params_total_b": null,
   "params_active_b": null,
   "moe": false,
   "params_note": "Card: MoE image-to-video, A14B. Total params not stated: unknown.",
   "params_source": "https://huggingface.co/Wan-AI/Wan2.2-I2V-A14B",
   "quant_format": "safetensors (as published)",
   "size_bytes": 126205359703,
   "size_gib": 117.54,
   "resident_bytes": 126205359703,
   "license": "apache-2.0",
   "abliterated": false,
   "public_folder": "usb-media/video",
   "status": "queued",
   "engine": "Wan2.2 generate.py / ComfyUI",
   "notes": "Card: MoE image-to-video, A14B. Total params not stated: unknown.",
   "context_native": null,
   "license_note": "Apache 2.0: commercial OK.",
   "non_commercial": false,
   "fit_note": "Same offload flags as T2V on the card.",
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "offload",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "offload",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "no",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": 130.2,
   "id": "wan2-2-i2v-a14b-safetensors"
  },
  {
   "repo": "tencent/HunyuanVideo-1.5",
   "hf_url": "https://huggingface.co/tencent/HunyuanVideo-1.5",
   "publisher": "tencent",
   "family": "HunyuanVideo 1.5",
   "modality": "text/image-to-video",
   "kind": "media",
   "params_total_b": 8.3,
   "params_active_b": null,
   "moe": false,
   "params_note": "Card: 8.3B parameters; repo is large because it ships several resolutions and components.",
   "params_source": "https://huggingface.co/tencent/HunyuanVideo-1.5",
   "quant_format": "safetensors (as published)",
   "size_bytes": 371771118874,
   "size_gib": 346.24,
   "resident_bytes": 371771118874,
   "license": "tencent-hunyuan-community",
   "abliterated": false,
   "public_folder": "usb-media/video",
   "status": "queued",
   "engine": "HunyuanVideo-1.5 repo / ComfyUI",
   "notes": "Card: 8.3B parameters; repo is large because it ships several resolutions and components.",
   "context_native": null,
   "license_note": "Tencent Hunyuan Community License: does NOT apply in the EU, UK or South Korea; >100M MAU needs a Tencent license.",
   "non_commercial": false,
   "fit_note": "Repo ships several resolutions/components (371 GB); one pipeline is loaded at a time. Card: 8.3B params, 'runs smoothly on consumer-grade GPUs'. Memory figure not stated on card: unknown.",
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "fits",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "offload",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "no",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "no",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "no",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": null,
   "id": "hunyuanvideo-1-5-safetensors"
  },
  {
   "repo": "black-forest-labs/flux-3-image-klein-comfy",
   "hf_url": "https://huggingface.co/black-forest-labs/flux-3-image-klein-comfy",
   "publisher": "black-forest-labs",
   "family": "FLUX.3 Klein (ComfyUI)",
   "modality": "image generation/editing",
   "kind": "media",
   "params_total_b": null,
   "params_active_b": null,
   "moe": false,
   "params_note": "Gated repo: card not readable without accepting terms. Params unknown.",
   "params_source": "https://huggingface.co/black-forest-labs/flux-3-image-klein-comfy",
   "quant_format": "unknown (gated)",
   "size_bytes": 93716186398,
   "size_gib": 87.28,
   "resident_bytes": 93716186398,
   "license": "flux-kommunity-license",
   "abliterated": false,
   "public_folder": "not downloaded",
   "status": "needs-Ian",
   "engine": "ComfyUI",
   "notes": "Gated: Ian must accept the terms on Hugging Face while signed in and provide a read token.",
   "context_native": null,
   "license_note": "Custom BFL license; read before any commercial use (gated).",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": null,
   "id": "flux-3-image-klein-comfy-unknown"
  },
  {
   "repo": "Lightricks/LTX-2.5",
   "hf_url": "https://huggingface.co/Lightricks/LTX-2.5",
   "publisher": "Lightricks",
   "family": "LTX-2.5",
   "modality": "image/text-to-video",
   "kind": "media",
   "params_total_b": null,
   "params_active_b": null,
   "moe": false,
   "params_note": "Gated repo: card not readable without accepting terms. Params unknown.",
   "params_source": "https://huggingface.co/Lightricks/LTX-2.5",
   "quant_format": "unknown (gated)",
   "size_bytes": 200854145597,
   "size_gib": 187.06,
   "resident_bytes": 200854145597,
   "license": "ltx-2.x-community-license-agreement",
   "abliterated": false,
   "public_folder": "not downloaded",
   "status": "needs-Ian",
   "engine": "ComfyUI",
   "notes": "Gated: Ian must accept the terms on Hugging Face while signed in and provide a read token.",
   "context_native": null,
   "license_note": "Custom Lightricks community license; read before commercial use (gated).",
   "non_commercial": false,
   "fit_note": null,
   "gguf_params_b": null,
   "fit": {
    "m5x1": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "m5x2": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "mbp": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "mini": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "omarchy": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "nas": {
     "verdict": "unknown",
     "est_tok_s": null
    },
    "neo": {
     "verdict": "unknown",
     "est_tok_s": null
    }
   },
   "est_memory_gb": null,
   "id": "ltx-2-5-unknown"
  }
 ],
 "on_device": {
  "generated": "2026-11-15 (America/Chicago)",
  "scanned": [
   "MacBook Pro (M4 Max)",
   "Mac mini (M4)"
  ],
  "not_scanned": [
   {
    "machine": "omarchy (AMD desktop)",
    "why": "SSH over Tailscale timed out on Nov 15"
   },
   {
    "machine": "MacBook Neo",
    "why": "offline on Nov 15"
   }
  ],
  "remote_only": [
   {
    "machine": "MacBook Pro (M4 Max)",
    "name": "Ollama minimax-m2:cloud",
    "note": "Ollama cloud link (manifest only, no local weights)"
   },
   {
    "machine": "MacBook Pro (M4 Max)",
    "name": "Ollama gpt-oss:120b-cloud",
    "note": "Ollama cloud link (manifest only, no local weights)"
   }
  ],
  "no_weights": [
   "MBP HF cache entries with refs only (nothing downloaded): ahmadw/YuE2-3B-MLX, mlx-community/bge-m3-mlx-8bit, mlx-community/gemma-4-26b-a4b-it-4bit, mlx-community/gemma-4-e4b-it-4bit, mlx-works/Ornith-1.5-35B-A3B-oQ4e-mtp; unsloth/gemma-4-26B-A4B-it (config files only, 27 KB). Mac mini: ahmadw/YuE2-3B-MLX, mlx-community/bge-m3-mlx-8bit (refs only).",
   "Mac mini ~/.hermes (Hermes agent install, 16 GB): no file over 1 GB; largest are its state database, backups and a git pack, none of them model weights. Not copied."
  ],
  "items": [
   {
    "id": "z-image-turbo",
    "name": "Z-Image Turbo",
    "repo": "Tongyi-MAI/Z-Image-Turbo",
    "format": "safetensors (Diffusers)",
    "size_bytes": 32848304893,
    "use": "image generation",
    "license": "apache-2.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "Weights live in the cache's shared blob store (hf-hub/blobs), copied alongside.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/Tongyi-MAI/Z-Image-Turbo",
    "size_gib": 30.59,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "pplx-computer-qwen38-27b-mlx",
    "name": "Perplexity Computer Qwen3.8-27B (MLX)",
    "repo": "perplexity-ai/pplx-computer-qwen-3-8-27b-mlx-20260824",
    "format": "MLX safetensors",
    "size_bytes": 16333696870,
    "use": "chat / vision LLM (image-text-to-text)",
    "license": "other",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "Card says license: other and names no license; read the card before commercial use. Base: Qwen/Qwen3.8-27B.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/perplexity-ai/pplx-computer-qwen-3-8-27b-mlx-20260824",
    "size_gib": 15.21,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "pplx-pii-masking-mlx",
    "name": "Perplexity PII masking (MLX)",
    "repo": "perplexity-ai/PII-Tracer-MLX",
    "format": "MLX safetensors",
    "size_bytes": 790891332,
    "use": "PII detection / masking (token classification)",
    "license": "mit",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "Downloaded as perplexity-ai/pplx-pii-masking-MLX; Hugging Face now redirects that name to perplexity-ai/PII-Tracer-MLX. LICENSE file in the snapshot is MIT.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/perplexity-ai/PII-Tracer-MLX",
    "size_gib": 0.74,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "qwen36-35b-a3b-mlx-3bit",
    "name": "Qwen3.6-35B-A3B UD MLX 3-bit",
    "repo": "unsloth/Qwen3.6-35B-A3B-UD-MLX-3bit",
    "format": "MLX safetensors (3-bit)",
    "size_bytes": 17392633006,
    "use": "chat / vision LLM",
    "license": "apache-2.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/unsloth/Qwen3.6-35B-A3B-UD-MLX-3bit",
    "size_gib": 16.2,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "qwen38-27b-ud-q4kxl",
    "name": "Qwen3.8-27B UD-Q4_K_XL + mmproj-F16",
    "repo": "unsloth/Qwen3.8-27B-GGUF",
    "format": "GGUF",
    "size_bytes": 18851002112,
    "use": "chat / vision LLM",
    "license": "apache-2.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "Not a duplicate: the NAS text archive holds the UD-Q8_K_XL file from this repo, not UD-Q4_K_XL or mmproj-F16, so these were copied. ~/models/Qwen3.8-27B-IQ4_XS.gguf on the MBP is only a symlink to this Q4_K_XL file.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF",
    "size_gib": 17.56,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "bge-small-en-gguf",
    "name": "bge-small-en v1.5 (GGUF f16)",
    "repo": "unsloth/bge-small-en-v1.5-GGUF",
    "format": "GGUF (f16)",
    "size_bytes": 67582560,
    "use": "embeddings",
    "license": "mit",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/unsloth/bge-small-en-v1.5-GGUF",
    "size_gib": 0.06,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "seedvr2-comfyui",
    "name": "SeedVR2 (ComfyUI)",
    "repo": "numz/SeedVR2_comfyUI",
    "format": "safetensors (fp16)",
    "size_bytes": 7284343622,
    "use": "video / image upscaling and restoration (video-to-video)",
    "license": "apache-2.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     },
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Same files on both Macs (names and byte sizes match); backed up once from the MacBook Pro. 3B fp16 model + VAE.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/numz/SeedVR2_comfyUI",
    "size_gib": 6.78,
    "macs": [
     "Mac mini (M4)",
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "ornith-15-9b-oq4e",
    "name": "Ornith 1.5 9B oQ4e-mtp (MLX)",
    "repo": "mlx-works/Ornith-1.5-9B-oQ4e-mtp",
    "format": "MLX safetensors (mixed 4/5-bit, oQ)",
    "size_bytes": 6245547422,
    "use": "chat LLM (served by oMLX)",
    "license": "unknown",
    "license_source": "Quant card states no license; base model ornith-ai/Ornith-1.5-9B is mit",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     },
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Four copies of the same weights (HF cache + oMLX folder on each Mac; names and sizes match). Backed up from the MacBook Pro only. Qwen3.5-9B based.",
    "hf_url": "https://huggingface.co/mlx-works/Ornith-1.5-9B-oQ4e-mtp",
    "size_gib": 5.82,
    "macs": [
     "Mac mini (M4)",
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "depth-anything-v2-small",
    "name": "Depth Anything V2 Small",
    "repo": "depth-anything/Depth-Anything-V2-Small-hf",
    "format": "safetensors",
    "size_bytes": 99175385,
    "use": "depth estimation",
    "license": "apache-2.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     },
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Identical on both Macs; backed up once.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/depth-anything/Depth-Anything-V2-Small-hf",
    "size_gib": 0.09,
    "macs": [
     "Mac mini (M4)",
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "claude-tokenizer",
    "name": "Claude tokenizer (Xenova)",
    "repo": "Xenova/claude-tokenizer",
    "format": "tokenizer.json (no weights)",
    "size_bytes": 1774213,
    "use": "token counting (tokenizer only, not a model)",
    "license": "mit",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     },
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/Xenova/claude-tokenizer",
    "size_gib": 0.0,
    "macs": [
     "Mac mini (M4)",
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "bge-m3-mlx-8bit",
    "name": "bge-m3 MLX 8-bit",
    "repo": "mlx-community/bge-m3-mlx-8bit",
    "format": "MLX safetensors (8-bit)",
    "size_bytes": 620779256,
    "use": "embeddings",
    "license": "mit",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     },
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Same weight files on both Macs; the MBP folder has 14 more small files (25 vs 11), so the MBP copy was backed up.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/mlx-community/bge-m3-mlx-8bit",
    "size_gib": 0.58,
    "macs": [
     "Mac mini (M4)",
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "gemma4-26b-a4b-mlx-4bit",
    "name": "Gemma 4 26B-A4B it MLX 4-bit",
    "repo": "mlx-community/gemma-4-26b-a4b-it-4bit",
    "format": "MLX safetensors (4-bit)",
    "size_bytes": 15373592206,
    "use": "chat / vision LLM",
    "license": "apache-2.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/mlx-community/gemma-4-26b-a4b-it-4bit",
    "size_gib": 14.32,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "flux-schnell-q8-local",
    "name": "FLUX schnell Q8 (local Diffusers save)",
    "repo": null,
    "format": "safetensors (Diffusers, 8-bit)",
    "size_bytes": 18000699218,
    "use": "image generation",
    "license": "unknown",
    "license_source": "Local conversion; the source repo isn't recorded in its files",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "Saved locally Oct 18 (flux-save.log). Name suggests FLUX.1-schnell, but the files don't say, so the license is left unknown.",
    "hf_url": null,
    "size_gib": 16.76,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "yue2-3b-mlx-8bit",
    "name": "YuE2 3B MLX 8-bit (local conversion)",
    "repo": null,
    "format": "MLX safetensors (8-bit) + MLX VAE",
    "size_bytes": 4532530702,
    "use": "music generation",
    "license": "cc-by-nc-4.0",
    "license_source": "Converted from m-a-p/YuE2-3B (per convert.py); that card is cc-by-nc-4.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     },
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Non-commercial. Same 8bit/ files on both Macs; backed up once with its convert/generate scripts. venv and outputs/ not copied.",
    "hf_url": null,
    "size_gib": 4.22,
    "macs": [
     "Mac mini (M4)",
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "qwen38-27b-dflash2",
    "name": "Qwen3.8-27B DFlash2 draft model",
    "repo": null,
    "format": "safetensors (bf16)",
    "size_bytes": 3848819135,
    "use": "speculative-decoding draft head for Qwen3.8-27B (experiment)",
    "license": "unknown",
    "license_source": "Local experiment; no license metadata",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "config.json architecture: DFlash2DraftModel.",
    "hf_url": null,
    "size_gib": 3.58,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "dflash2-v2-gguf",
    "name": "DFlash2 v2 Q4_K_M self16 (GGUF)",
    "repo": null,
    "format": "GGUF",
    "size_bytes": 1327654720,
    "use": "speculative-decoding draft (experiment)",
    "license": "unknown",
    "license_source": "GGUF metadata has no license field",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "GGUF architecture: dflash, name qwen38-27b-dflash2.",
    "hf_url": null,
    "size_gib": 1.24,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "qwen38-27b-tail-iq4xs",
    "name": "Qwen3.8-27B tail L=40 IQ4_XS (partial GGUF)",
    "repo": null,
    "format": "GGUF (IQ4_XS, partial)",
    "size_bytes": 7719767296,
    "use": "experiment: layers from 40 on, no output head",
    "license": "apache-2.0",
    "license_source": "GGUF metadata (base Qwen/Qwen3.8-27B, quantized by Unsloth)",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "notes": "Not a standalone runnable model.",
    "hf_url": null,
    "size_gib": 7.19,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "whisper-tiny-en",
    "name": "Whisper tiny.en (openai-whisper)",
    "repo": "openai/whisper-tiny.en",
    "format": "PyTorch .pt",
    "size_bytes": 75571315,
    "use": "speech-to-text",
    "license": "apache-2.0",
    "locations": [
     {
      "machine": "MacBook Pro (M4 Max)"
     }
    ],
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/openai/whisper-tiny.en",
    "size_gib": 0.07,
    "macs": [
     "MacBook Pro (M4 Max)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "black-forest-labs/FLUX.1-dev",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "flux1-dev",
    "name": "FLUX.1 dev",
    "format": "safetensors (Diffusers)",
    "size_bytes": 33746382743,
    "use": "image generation",
    "license": "flux-1-dev-non-commercial-license",
    "notes": "Non-commercial.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/black-forest-labs/FLUX.1-dev",
    "size_gib": 31.43,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "black-forest-labs/FLUX.1-schnell",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "flux1-schnell",
    "name": "FLUX.1 schnell",
    "format": "safetensors (Diffusers)",
    "size_bytes": 33725922192,
    "use": "image generation",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/black-forest-labs/FLUX.1-schnell",
    "size_gib": 31.41,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "mlx-community/Qwen3-TTS-12Hz-1.7B-CustomVoice-8bit",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "qwen3-tts-1.7b-cv-8bit",
    "name": "Qwen3-TTS 1.7B CustomVoice MLX 8-bit",
    "format": "MLX safetensors (8-bit)",
    "size_bytes": 3080141538,
    "use": "text-to-speech",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/mlx-community/Qwen3-TTS-12Hz-1.7B-CustomVoice-8bit",
    "size_gib": 2.87,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "mlx-community/Qwen3-TTS-12Hz-0.6B-CustomVoice-8bit",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "qwen3-tts-0.6b-cv-8bit",
    "name": "Qwen3-TTS 0.6B CustomVoice MLX 8-bit",
    "format": "MLX safetensors (8-bit)",
    "size_bytes": 1973572801,
    "use": "text-to-speech",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/mlx-community/Qwen3-TTS-12Hz-0.6B-CustomVoice-8bit",
    "size_gib": 1.84,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "mlx-community/Qwen3.5-4B-MLX-4bit",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "qwen35-4b-mlx-4bit",
    "name": "Qwen3.5 4B MLX 4-bit",
    "format": "MLX safetensors (4-bit)",
    "size_bytes": 3061132920,
    "use": "chat LLM",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/mlx-community/Qwen3.5-4B-MLX-4bit",
    "size_gib": 2.85,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "mlx-community/Qwen3.5-2B-MLX-4bit",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "qwen35-2b-mlx-4bit",
    "name": "Qwen3.5 2B MLX 4-bit",
    "format": "MLX safetensors (4-bit)",
    "size_bytes": 1749083534,
    "use": "chat LLM",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/mlx-community/Qwen3.5-2B-MLX-4bit",
    "size_gib": 1.63,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "facebook/w2v-bert-2.0",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "w2v-bert-2",
    "name": "w2v-BERT 2.0",
    "format": "safetensors",
    "size_bytes": 2322065885,
    "use": "speech encoder (audio feature extraction)",
    "license": "mit",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/facebook/w2v-bert-2.0",
    "size_gib": 2.16,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "neuphonic/neucodec",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "neucodec",
    "name": "NeuCodec",
    "format": "PyTorch .bin",
    "size_bytes": 2321024854,
    "use": "neural audio codec (used by NeuTTS)",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/neuphonic/neucodec",
    "size_gib": 2.16,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "neuphonic/neutts-air-q4-gguf",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "neutts-air-q4",
    "name": "NeuTTS Air Q4 (GGUF)",
    "format": "GGUF (Q4_0)",
    "size_bytes": 527054304,
    "use": "text-to-speech",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/neuphonic/neutts-air-q4-gguf",
    "size_gib": 0.49,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "google/embeddinggemma-2",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "embeddinggemma-2",
    "name": "EmbeddingGemma 2",
    "format": "safetensors",
    "size_bytes": 1521119809,
    "use": "embeddings",
    "license": "apache-2.0",
    "notes": "~/models/embeddinggemma-2 holds only demo scripts; the weights are in the HF cache.",
    "license_source": "Hugging Face card metadata",
    "hf_url": "https://huggingface.co/google/embeddinggemma-2",
    "size_gib": 1.42,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "laion/CLIP-ViT-B-32-laion2B-s34B-b79K",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "clip-vit-b32-laion2b",
    "name": "CLIP ViT-B/32 (LAION-2B)",
    "format": "safetensors (open_clip)",
    "size_bytes": 605143316,
    "use": "image-text embeddings / zero-shot image classification",
    "license": "mit",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/laion/CLIP-ViT-B-32-laion2B-s34B-b79K",
    "size_gib": 0.56,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "hexgrad/Kokoro-82M",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "kokoro-82m",
    "name": "Kokoro 82M",
    "format": "PyTorch .pth",
    "size_bytes": 327738002,
    "use": "text-to-speech",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/hexgrad/Kokoro-82M",
    "size_gib": 0.31,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "mlx-community/Kokoro-82M-8bit",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "kokoro-82m-mlx-8bit",
    "name": "Kokoro 82M MLX 8-bit",
    "format": "MLX safetensors (8-bit)",
    "size_bytes": 317533045,
    "use": "text-to-speech",
    "license": "apache-2.0",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/mlx-community/Kokoro-82M-8bit",
    "size_gib": 0.3,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "repo": "Systran/faster-whisper-base",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "id": "faster-whisper-base",
    "name": "faster-whisper base",
    "format": "CTranslate2",
    "size_bytes": 147882941,
    "use": "speech-to-text",
    "license": "mit",
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/Systran/faster-whisper-base",
    "size_gib": 0.14,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "gemma4-e4b-mlx-4bit",
    "name": "Gemma 4 E4B it MLX 4-bit",
    "repo": "mlx-community/gemma-4-e4b-it-4bit",
    "format": "MLX safetensors (4-bit)",
    "size_bytes": 5179241512,
    "use": "chat / vision LLM",
    "license": "gemma",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "license_source": "Hugging Face card metadata",
    "notes": "",
    "hf_url": "https://huggingface.co/mlx-community/gemma-4-e4b-it-4bit",
    "size_gib": 4.82,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "mellum21-12b-thinking-q4km",
    "name": "Mellum2.1 12B-A2.5B Thinking Q4_K_M",
    "repo": null,
    "format": "GGUF (Q4_K_M)",
    "size_bytes": 8071295264,
    "use": "LLM (JetBrains Mellum, thinking variant)",
    "license": "apache-2.0",
    "license_source": "GGUF metadata (base JetBrains/Mellum2-12B-A2.5B-Base, whose card is also apache-2.0)",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "MoE (GGUF size label 64x1.2B).",
    "hf_url": null,
    "size_gib": 7.52,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "ollama-gemma4-e4b",
    "name": "Ollama gemma4:e4b",
    "repo": null,
    "format": "Ollama (GGUF blobs)",
    "size_bytes": 9608350718,
    "use": "chat LLM",
    "license": "Apache License",
    "license_source": "First line of the model's Ollama license layer",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Whole Ollama store copied (blobs + manifests), so it can be restored by copying back to ~/.ollama/models.",
    "hf_url": null,
    "size_gib": 8.95,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "ollama-qwen35-9b",
    "name": "Ollama qwen3.5:9b",
    "repo": null,
    "format": "Ollama (GGUF blobs)",
    "size_bytes": 6594474711,
    "use": "chat LLM",
    "license": "Apache License",
    "license_source": "First line of the model's Ollama license layer",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Whole Ollama store copied (blobs + manifests), so it can be restored by copying back to ~/.ollama/models.",
    "hf_url": null,
    "size_gib": 6.14,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "ollama-qwen3-8b",
    "name": "Ollama qwen3:8b",
    "repo": null,
    "format": "Ollama (GGUF blobs)",
    "size_bytes": 5225388164,
    "use": "chat LLM",
    "license": "Apache License",
    "license_source": "First line of the model's Ollama license layer",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Whole Ollama store copied (blobs + manifests), so it can be restored by copying back to ~/.ollama/models.",
    "hf_url": null,
    "size_gib": 4.87,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "ollama-gemma3-4b",
    "name": "Ollama gemma3:4b",
    "repo": null,
    "format": "Ollama (GGUF blobs)",
    "size_bytes": 3338801804,
    "use": "chat LLM",
    "license": "Gemma Terms of Use",
    "license_source": "First line of the model's Ollama license layer",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Whole Ollama store copied (blobs + manifests), so it can be restored by copying back to ~/.ollama/models.",
    "hf_url": null,
    "size_gib": 3.11,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   },
   {
    "id": "ollama-qwen3-4b",
    "name": "Ollama qwen3:4b",
    "repo": null,
    "format": "Ollama (GGUF blobs)",
    "size_bytes": 2497293931,
    "use": "chat LLM",
    "license": "Apache License",
    "license_source": "First line of the model's Ollama license layer",
    "locations": [
     {
      "machine": "Mac mini (M4)"
     }
    ],
    "notes": "Whole Ollama store copied (blobs + manifests), so it can be restored by copying back to ~/.ollama/models.",
    "hf_url": null,
    "size_gib": 2.33,
    "macs": [
     "Mac mini (M4)"
    ],
    "backup_status": "copying"
   }
  ],
  "total_unique_bytes": 277355009251
 }
}