data.json

{
 "meta": {
  "gpu": "Radeon 8060S Graphics",
  "arch": "gfx1151",
  "vram_total": 64.0,
  "rocm": "7.13.26162",
  "kernel": "7.0.14-101.fc43.x86_64",
  "torch": "2.11.0+rocm7.13.0a20260425",
  "n_rows": 44,
  "n_models": 11
 },
 "rows": [
  {
   "label": "ornith-1.0-35b-Q4_K_M",
   "family": "ornith-1.0-35b",
   "quant": "Q4_K_M",
   "arch": "MoE",
   "vram": 20.1,
   "size_gb": 19.7,
   "decode": 74.1,
   "decode_d4k": 69.9,
   "prefill": 1102.5,
   "prefill_4k": 1060.2
  },
  {
   "label": "Qwen3.6-35B-A3B-UD-Q4_K_M",
   "family": "Qwen3.6-35B-A3B",
   "quant": "UD-Q4_K_M",
   "arch": "MoE",
   "vram": 20.7,
   "size_gb": 20.6,
   "decode": 62.2,
   "decode_d4k": 59.8,
   "prefill": 1014.3,
   "prefill_4k": 1057.1
  },
  {
   "label": "gpt-oss-120b-Q4_K_M",
   "family": "gpt-oss-120b",
   "quant": "Q4_K_M",
   "arch": "MoE",
   "vram": 58.9,
   "size_gb": 46.2,
   "decode": 57.2,
   "decode_d4k": 56.2,
   "prefill": 648.4,
   "prefill_4k": 653.7
  },
  {
   "label": "Qwen3.6-35B-A3B-Q8_0",
   "family": "Qwen3.6-35B-A3B",
   "quant": "Q8_0",
   "arch": "MoE",
   "vram": 34.7,
   "size_gb": 34.4,
   "decode": 53.5,
   "decode_d4k": 50.9,
   "prefill": 1136.0,
   "prefill_4k": 1111.3
  },
  {
   "label": "Qwen3.6-27B-Q3_K_M",
   "family": "Qwen3.6-27B",
   "quant": "Q3_K_M",
   "arch": "dense",
   "vram": 13.2,
   "size_gb": 12.7,
   "decode": 14.4,
   "decode_d4k": 14.6,
   "prefill": 297.0,
   "prefill_4k": 305.8
  },
  {
   "label": "Qwen3.6-27B-Q4_0",
   "family": "Qwen3.6-27B",
   "quant": "Q4_0",
   "arch": "dense",
   "vram": 14.9,
   "size_gb": 14.7,
   "decode": 13.4,
   "decode_d4k": 13.1,
   "prefill": 339.4,
   "prefill_4k": 327.8
  },
  {
   "label": "Qwen3.8-27B-UD-Q4_K_XL",
   "family": "Qwen3.8-27B",
   "quant": "UD-Q4_K_XL",
   "arch": "dense",
   "vram": 16.9,
   "size_gb": 16.7,
   "decode": 11.9,
   "decode_d4k": 11.7,
   "prefill": 349.0,
   "prefill_4k": 335.5
  },
  {
   "label": "cognitivecomputations_Dolphin-Mistral-24B-Venice-Edition-Q6_K",
   "family": "cognitivecomputations_Dolphin-Mistral-24B-Venice-Edition",
   "quant": "Q6_K",
   "arch": "dense",
   "vram": 18.4,
   "size_gb": 18.0,
   "decode": 11.5,
   "decode_d4k": 11.2,
   "prefill": 380.1,
   "prefill_4k": 353.5
  },
  {
   "label": "gemma-4-31B-it-UD-Q4_K_XL",
   "family": "gemma-4-31B-it",
   "quant": "UD-Q4_K_XL",
   "arch": "dense",
   "vram": 19.6,
   "size_gb": 17.5,
   "decode": 11.2,
   "decode_d4k": 10.2,
   "prefill": 279.7,
   "prefill_4k": 261.9
  },
  {
   "label": "Qwen3.6-27B-Q5_K_M",
   "family": "Qwen3.6-27B",
   "quant": "Q5_K_M",
   "arch": "dense",
   "vram": 18.5,
   "size_gb": 18.2,
   "decode": 11.0,
   "decode_d4k": 10.8,
   "prefill": 330.6,
   "prefill_4k": 325.2
  },
  {
   "label": "Qwen3.6-27B-Q6_K",
   "family": "Qwen3.6-27B",
   "quant": "Q6_K",
   "arch": "dense",
   "vram": 20.9,
   "size_gb": 21.0,
   "decode": 9.5,
   "decode_d4k": 9.4,
   "prefill": 303.4,
   "prefill_4k": 294.3
  }
 ]
}