Skip to main content

MODELS_JSON

Constant MODELS_JSON 

Source
pub const MODELS_JSON: &str = "{\n  \"schemaVersion\": 2,\n  \"generatedAt\": \"2026-09-01T02:44:54.336Z\",\n  \"models\": [\n    {\n      \"id\": \"gemma4:e4b-it-qat\",\n      \"family\": \"gemma4\",\n      \"params\": \"8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"vision\",\n        \"code\",\n        \"tools\",\n        \"reasoning\"\n      ],\n      \"releaseDate\": \"2026-03-31\",\n      \"source\": {\n        \"ollama\": \"gemma4:e4b-it-qat\",\n        \"hf\": \"google/gemma-4-E4B-it-qat-q4_0-gguf\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_0\",\n          \"diskBytes\": 5154939136,\n          \"minRamBytes\": 5928180007,\n          \"minVramBytes\": 5928180007,\n          \"sha256\": \"e8b6a059ba86947a44ace84d6e5679795bc41862c25c30513142588f0e9dba1d\"\n        }\n      ],\n      \"benchmarkProxy\": 0.52\n    },\n    {\n      \"id\": \"kimi-k2-thinking\",\n      \"family\": \"kimi-k2-thinking\",\n      \"params\": \"1T\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"32B\",\n      \"license\": \"modified-mit\",\n      \"openWeight\": true,\n      \"contextLength\": 262144,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2025-11-06\",\n      \"source\": {\n        \"hf\": \"moonshotai/Kimi-K2-Thinking\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 600000000000,\n          \"minRamBytes\": 690000000000,\n          \"minVramBytes\": 690000000000\n        }\n      ],\n      \"benchmarkProxy\": 0.99\n    },\n    {\n      \"id\": \"kimi-linear\",\n      \"family\": \"kimi-linear\",\n      \"params\": \"48B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"3B\",\n      \"license\": \"modified-mit\",\n      \"openWeight\": true,\n      \"contextLength\": 262144,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\"\n      ],\n      \"releaseDate\": \"2025-10-30\",\n      \"source\": {\n        \"hf\": \"moonshotai/Kimi-Linear-48B-A3B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 29000000000,\n          \"minRamBytes\": 33350000000,\n          \"minVramBytes\": 33350000000\n        }\n      ],\n      \"benchmarkProxy\": 0.78\n    },\n    {\n      \"id\": \"kimi-k2:base\",\n      \"family\": \"kimi-k2\",\n      \"params\": \"1T\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"32B\",\n      \"license\": \"modified-mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-07-11\",\n      \"source\": {\n        \"hf\": \"moonshotai/Kimi-K2-Base\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 600000000000,\n          \"minRamBytes\": 690000000000,\n          \"minVramBytes\": 690000000000\n        }\n      ],\n      \"benchmarkProxy\": 0.97\n    },\n    {\n      \"id\": \"kimi-k2:instruct\",\n      \"family\": \"kimi-k2\",\n      \"params\": \"1T\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"32B\",\n      \"license\": \"modified-mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-07-11\",\n      \"source\": {\n        \"ollama\": \"kimi-k2:instruct\",\n        \"hf\": \"moonshotai/Kimi-K2-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 600000000000,\n          \"minRamBytes\": 690000000000,\n          \"minVramBytes\": 690000000000\n        }\n      ],\n      \"benchmarkProxy\": 0.97\n    },\n    {\n      \"id\": \"gemma3n:e2b\",\n      \"family\": \"gemma3n\",\n      \"params\": \"6B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\",\n        \"vision\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-06-26\",\n      \"source\": {\n        \"ollama\": \"gemma3n:e2b\",\n        \"hf\": \"google/gemma-3n-E2B-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 5621607424,\n          \"minRamBytes\": 6464848538,\n          \"minVramBytes\": 6464848538,\n          \"sha256\": \"3839a254cf2d00b208c6e2524c129e4438f9d106bba4c3fbc12b631f519d1de1\"\n        }\n      ],\n      \"benchmarkProxy\": 0.42\n    },\n    {\n      \"id\": \"gemma3n:e4b\",\n      \"family\": \"gemma3n\",\n      \"params\": \"8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\",\n        \"vision\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-06-26\",\n      \"source\": {\n        \"ollama\": \"gemma3n:e4b\",\n        \"hf\": \"google/gemma-3n-E4B-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 7547579904,\n          \"minRamBytes\": 8679716890,\n          \"minVramBytes\": 8679716890,\n          \"sha256\": \"38e8dcc30df4eb0e29eaf5c74ba6ce3f2cd66badad50768fc14362acfb8b8cb6\"\n        }\n      ],\n      \"benchmarkProxy\": 0.5\n    },\n    {\n      \"id\": \"kimi-dev-72b\",\n      \"family\": \"kimi-dev\",\n      \"params\": \"72B\",\n      \"architecture\": \"dense\",\n      \"license\": \"modified-mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"code\",\n        \"chat\",\n        \"reasoning\"\n      ],\n      \"releaseDate\": \"2025-06-16\",\n      \"source\": {\n        \"hf\": \"moonshotai/Kimi-Dev-72B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 43000000000,\n          \"minRamBytes\": 49450000000,\n          \"minVramBytes\": 49450000000\n        }\n      ],\n      \"benchmarkProxy\": 0.78\n    },\n    {\n      \"id\": \"qwen3:0.6b\",\n      \"family\": \"qwen3\",\n      \"params\": \"0.6B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 40960,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-29\",\n      \"source\": {\n        \"ollama\": \"qwen3:0.6b\",\n        \"hf\": \"Qwen/Qwen3-0.6B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 522640096,\n          \"minRamBytes\": 601036111,\n          \"minVramBytes\": 601036111,\n          \"sha256\": \"7f4030143c1c477224c5434f8272c662a8b042079a0a584f0a27a1684fe2e1fa\"\n        }\n      ],\n      \"benchmarkProxy\": 0.22\n    },\n    {\n      \"id\": \"qwen3:1.7b\",\n      \"family\": \"qwen3\",\n      \"params\": \"1.7B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 40960,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-29\",\n      \"source\": {\n        \"ollama\": \"qwen3:1.7b\",\n        \"hf\": \"Qwen/Qwen3-1.7B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 1359279776,\n          \"minRamBytes\": 1563171743,\n          \"minVramBytes\": 1563171743,\n          \"sha256\": \"3d0b790534fe4b79525fc3692950408dca41171676ed7e21db57af5c65ef6ab6\"\n        }\n      ],\n      \"benchmarkProxy\": 0.3\n    },\n    {\n      \"id\": \"qwen3:4b\",\n      \"family\": \"qwen3\",\n      \"params\": \"4B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 262144,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-29\",\n      \"source\": {\n        \"ollama\": \"qwen3:4b\",\n        \"hf\": \"Qwen/Qwen3-4B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 2497280480,\n          \"minRamBytes\": 2871872552,\n          \"minVramBytes\": 2871872552,\n          \"sha256\": \"3e4cb14174460404e7a233e531675303b2fbf7749c02f91864fe311ab6344e4f\"\n        }\n      ],\n      \"benchmarkProxy\": 0.46\n    },\n    {\n      \"id\": \"qwen3:14b\",\n      \"family\": \"qwen3\",\n      \"params\": \"14B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 40960,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-28\",\n      \"source\": {\n        \"ollama\": \"qwen3:14b\",\n        \"hf\": \"Qwen/Qwen3-14B\",\n        \"gguf\": {\n          \"repo\": \"Qwen/Qwen3-14B-GGUF\",\n          \"revision\": \"530227a7d994db8eca5ab5ced2fb692b614357fd\",\n          \"file\": \"Qwen3-14B-Q4_K_M.gguf\",\n          \"sha256\": \"500a8806e85ee9c83f3ae08420295592451379b4f8cf2d0f41c15dffeb6b81f0\"\n        }\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 9276184896,\n          \"minRamBytes\": 10667612631,\n          \"minVramBytes\": 10667612631,\n          \"sha256\": \"a8cc1361f3145dc01f6d77c6c82c9116b9ffe3c97b34716fe20418455876c40e\"\n        }\n      ],\n      \"benchmarkProxy\": 0.62\n    },\n    {\n      \"id\": \"qwen3:235b-a22b\",\n      \"family\": \"qwen3\",\n      \"params\": \"235B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"22B\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 262144,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-28\",\n      \"source\": {\n        \"ollama\": \"qwen3:235b-a22b\",\n        \"hf\": \"Qwen/Qwen3-235B-A22B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 142154073664,\n          \"minRamBytes\": 163477184714,\n          \"minVramBytes\": 163477184714,\n          \"sha256\": \"791d5d11998e006548d6b58c31756562ea61446ebc7d19686608402a797ecc82\"\n        }\n      ],\n      \"benchmarkProxy\": 0.95\n    },\n    {\n      \"id\": \"qwen3:30b-a3b\",\n      \"family\": \"qwen3\",\n      \"params\": \"30B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"3B\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 262144,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-28\",\n      \"source\": {\n        \"ollama\": \"qwen3:30b-a3b\",\n        \"hf\": \"Qwen/Qwen3-30B-A3B\",\n        \"gguf\": {\n          \"repo\": \"Qwen/Qwen3-30B-A3B-GGUF\",\n          \"revision\": \"e4d4bafdfb96a411a163846265362aceb0b9c63a\",\n          \"file\": \"Qwen3-30B-A3B-Q4_K_M.gguf\",\n          \"sha256\": \"0d003f6662faee786ed5da3e31b29c978de5ae5d275c8794c606a7f3c01aa8f5\"\n        }\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 18556685856,\n          \"minRamBytes\": 21340188735,\n          \"minVramBytes\": 21340188735,\n          \"sha256\": \"58574f2e94b99fb9e4391408b57e5aeaaaec10f6384e9a699fc2cb43a5c8eabf\"\n        }\n      ],\n      \"benchmarkProxy\": 0.74\n    },\n    {\n      \"id\": \"qwen3:32b\",\n      \"family\": \"qwen3\",\n      \"params\": \"32B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 40960,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-28\",\n      \"source\": {\n        \"ollama\": \"qwen3:32b\",\n        \"hf\": \"Qwen/Qwen3-32B\",\n        \"gguf\": {\n          \"repo\": \"Qwen/Qwen3-32B-GGUF\",\n          \"revision\": \"938a7432affaec9157f883a87164e2646ae17555\",\n          \"file\": \"Qwen3-32B-Q4_K_M.gguf\",\n          \"sha256\": \"efd971561896866f0e910cce52761ca77b1b138090c7f15fe284676d57d1f689\"\n        }\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 20201240160,\n          \"minRamBytes\": 23231426184,\n          \"minVramBytes\": 23231426184,\n          \"sha256\": \"3291abe70f16ee9682de7bfae08db5373ea9d6497e614aaad63340ad421d6312\"\n        }\n      ],\n      \"benchmarkProxy\": 0.74\n    },\n    {\n      \"id\": \"qwen3:8b\",\n      \"family\": \"qwen3\",\n      \"params\": \"8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 40960,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-04-28\",\n      \"source\": {\n        \"ollama\": \"qwen3:8b\",\n        \"hf\": \"Qwen/Qwen3-8B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 5225374496,\n          \"minRamBytes\": 6009180671,\n          \"minVramBytes\": 6009180671,\n          \"sha256\": \"a3de86cd1c132c822487ededd47a324c50491393e6565cd14bafa40d0b8e686f\"\n        }\n      ],\n      \"benchmarkProxy\": 0.54\n    },\n    {\n      \"id\": \"glm4:9b\",\n      \"family\": \"glm4\",\n      \"params\": \"9B\",\n      \"architecture\": \"dense\",\n      \"license\": \"mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2025-04-14\",\n      \"source\": {\n        \"ollama\": \"glm4:9b\",\n        \"hf\": \"THUDM/GLM-4-9B-0414\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 5455319040,\n          \"minRamBytes\": 6273616896,\n          \"minVramBytes\": 6273616896,\n          \"sha256\": \"b506a070d1152798d435ec4e7687336567ae653b3106f73b7b4ac7be1cbc4449\"\n        }\n      ],\n      \"benchmarkProxy\": 0.49\n    },\n    {\n      \"id\": \"kimi-vl-a3b\",\n      \"family\": \"kimi-vl\",\n      \"params\": \"16B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"3B\",\n      \"license\": \"modified-mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"vision\"\n      ],\n      \"releaseDate\": \"2025-04-10\",\n      \"source\": {\n        \"hf\": \"moonshotai/Kimi-VL-A3B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 10000000000,\n          \"minRamBytes\": 11500000000,\n          \"minVramBytes\": 11500000000\n        }\n      ],\n      \"benchmarkProxy\": 0.64\n    },\n    {\n      \"id\": \"gemma3:12b\",\n      \"family\": \"gemma3\",\n      \"params\": \"12B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"vision\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-03-12\",\n      \"source\": {\n        \"ollama\": \"gemma3:12b\",\n        \"hf\": \"google/gemma-3-12b-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 8149180896,\n          \"minRamBytes\": 9371558031,\n          \"minVramBytes\": 9371558031,\n          \"sha256\": \"e8ad13eff07a78d89926e9e8b882317d082ef5bf9768ad7b50fcdbbcd63748de\"\n        }\n      ],\n      \"benchmarkProxy\": 0.58\n    },\n    {\n      \"id\": \"gemma3:27b\",\n      \"family\": \"gemma3\",\n      \"params\": \"27B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"vision\",\n        \"code\",\n        \"reasoning\"\n      ],\n      \"releaseDate\": \"2025-03-12\",\n      \"source\": {\n        \"ollama\": \"gemma3:27b\",\n        \"hf\": \"google/gemma-3-27b-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 17396927584,\n          \"minRamBytes\": 20006466722,\n          \"minVramBytes\": 20006466722,\n          \"sha256\": \"e796792eba26c4d3b04b0ac5adb01a453dd9ec2dfd83b6c59cbf6fe5f30b0f68\"\n        }\n      ],\n      \"benchmarkProxy\": 0.7\n    },\n    {\n      \"id\": \"gemma3:4b\",\n      \"family\": \"gemma3\",\n      \"params\": \"4B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"vision\"\n      ],\n      \"releaseDate\": \"2025-03-12\",\n      \"source\": {\n        \"ollama\": \"gemma3:4b\",\n        \"hf\": \"google/gemma-3-4b-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 3338792448,\n          \"minRamBytes\": 3839611316,\n          \"minVramBytes\": 3839611316,\n          \"sha256\": \"aeda25e63ebd698fab8638ffb778e68bed908b960d39d0becc650fa981609d25\"\n        }\n      ],\n      \"benchmarkProxy\": 0.42\n    },\n    {\n      \"id\": \"phi4-mini:3.8b\",\n      \"family\": \"phi4-mini\",\n      \"params\": \"3.8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"tools\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-02-26\",\n      \"source\": {\n        \"ollama\": \"phi4-mini:3.8b\",\n        \"hf\": \"microsoft/Phi-4-mini-instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 2491874624,\n          \"minRamBytes\": 2865655818,\n          \"minVramBytes\": 2865655818,\n          \"sha256\": \"3c168af1dea0a414299c7d9077e100ac763370e5a98b3c53801a958a47f0a5db\"\n        }\n      ],\n      \"benchmarkProxy\": 0.38\n    },\n    {\n      \"id\": \"mistral-small:24b\",\n      \"family\": \"mistral-small\",\n      \"params\": \"24B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2025-01-30\",\n      \"source\": {\n        \"ollama\": \"mistral-small:24b\",\n        \"hf\": \"mistralai/Mistral-Small-24B-Instruct-2501\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 14333908384,\n          \"minRamBytes\": 16483994642,\n          \"minVramBytes\": 16483994642,\n          \"sha256\": \"102a747c137683e81d431dab05d8f2158df4ab6f162f8f9019425a43d51e0e9f\"\n        }\n      ],\n      \"kvBytesPerToken\": 163840,\n      \"benchmarkProxy\": 0.63\n    },\n    {\n      \"id\": \"deepseek-r1:14b\",\n      \"family\": \"deepseek-r1\",\n      \"params\": \"14B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-01-20\",\n      \"source\": {\n        \"ollama\": \"deepseek-r1:14b\",\n        \"hf\": \"deepseek-ai/DeepSeek-R1-Distill-Qwen-14B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 8988109952,\n          \"minRamBytes\": 10336326445,\n          \"minVramBytes\": 10336326445,\n          \"sha256\": \"6e9f90f02bb3b39b59e81916e8cfce9deb45aeaeb9a54a5be4414486b907dc1e\"\n        }\n      ],\n      \"benchmarkProxy\": 0.64\n    },\n    {\n      \"id\": \"deepseek-r1:32b\",\n      \"family\": \"deepseek-r1\",\n      \"params\": \"32B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-01-20\",\n      \"source\": {\n        \"ollama\": \"deepseek-r1:32b\",\n        \"hf\": \"deepseek-ai/DeepSeek-R1-Distill-Qwen-32B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 19851335552,\n          \"minRamBytes\": 22829035885,\n          \"minVramBytes\": 22829035885,\n          \"sha256\": \"6150cb382311b69f09cc0f9a1b69fc029cbd742b66bb8ec531aa5ecf5c613e93\"\n        }\n      ],\n      \"benchmarkProxy\": 0.76\n    },\n    {\n      \"id\": \"deepseek-r1:671b\",\n      \"family\": \"deepseek-r1\",\n      \"params\": \"671B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"37B\",\n      \"license\": \"mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"reasoning\",\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-01-20\",\n      \"source\": {\n        \"ollama\": \"deepseek-r1:671b\",\n        \"hf\": \"deepseek-ai/DeepSeek-R1\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 404430186848,\n          \"minRamBytes\": 465094714876,\n          \"minVramBytes\": 465094714876,\n          \"sha256\": \"439dd1a5e05286918f54941f49f9b56118c757440f6333f67f1cd5cbb5c8520b\"\n        }\n      ],\n      \"benchmarkProxy\": 0.97\n    },\n    {\n      \"id\": \"deepseek-r1:70b\",\n      \"family\": \"deepseek-r1\",\n      \"params\": \"70B\",\n      \"architecture\": \"dense\",\n      \"license\": \"llama-3.3-community\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-01-20\",\n      \"source\": {\n        \"ollama\": \"deepseek-r1:70b\",\n        \"hf\": \"deepseek-ai/DeepSeek-R1-Distill-Llama-70B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 42520395616,\n          \"minRamBytes\": 48898454959,\n          \"minVramBytes\": 48898454959,\n          \"sha256\": \"4cd576d9aa16961244012223abf01445567b061f1814b57dfef699e4cf8df339\"\n        }\n      ],\n      \"benchmarkProxy\": 0.82\n    },\n    {\n      \"id\": \"deepseek-r1:7b\",\n      \"family\": \"deepseek-r1\",\n      \"params\": \"7B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-01-20\",\n      \"source\": {\n        \"ollama\": \"deepseek-r1:7b\",\n        \"hf\": \"deepseek-ai/DeepSeek-R1-Distill-Qwen-7B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 4683073184,\n          \"minRamBytes\": 5385534162,\n          \"minVramBytes\": 5385534162,\n          \"sha256\": \"96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49\"\n        }\n      ],\n      \"benchmarkProxy\": 0.56\n    },\n    {\n      \"id\": \"deepseek-r1:8b\",\n      \"family\": \"deepseek-r1\",\n      \"params\": \"8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"llama-3.1-community\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"reasoning\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2025-01-20\",\n      \"source\": {\n        \"ollama\": \"deepseek-r1:8b\",\n        \"hf\": \"deepseek-ai/DeepSeek-R1-Distill-Llama-8B\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 5225373760,\n          \"minRamBytes\": 6009179824,\n          \"minVramBytes\": 6009179824,\n          \"sha256\": \"e6a7edc1a4d7d9b2de136a221a57336b76316cfe53a252aeba814496c5ae439d\"\n        }\n      ],\n      \"benchmarkProxy\": 0.56\n    },\n    {\n      \"id\": \"deepseek-v3\",\n      \"family\": \"deepseek-v3\",\n      \"params\": \"671B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"37B\",\n      \"license\": \"deepseek\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-12-26\",\n      \"source\": {\n        \"ollama\": \"deepseek-v3\",\n        \"hf\": \"deepseek-ai/DeepSeek-V3\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 404430186016,\n          \"minRamBytes\": 465094713919,\n          \"minVramBytes\": 465094713919,\n          \"sha256\": \"d83c18fb2a2ccca56d641920f01e6fe533dfb3fcf4b77c007c931497cd24a517\"\n        }\n      ],\n      \"benchmarkProxy\": 0.95\n    },\n    {\n      \"id\": \"granite3.1:2b\",\n      \"family\": \"granite3.1\",\n      \"params\": \"2B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-12-18\",\n      \"source\": {\n        \"ollama\": \"granite3.1-dense:2b\",\n        \"hf\": \"ibm-granite/granite-3.1-2b-instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 1569676320,\n          \"minRamBytes\": 1805127768,\n          \"minVramBytes\": 1805127768,\n          \"sha256\": \"5c56bb0256a2c402e95282a29bb5cb747bb805eda0e14a84b1f6c594a297ec1a\"\n        }\n      ],\n      \"benchmarkProxy\": 0.31\n    },\n    {\n      \"id\": \"granite3.1:8b\",\n      \"family\": \"granite3.1\",\n      \"params\": \"8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-12-18\",\n      \"source\": {\n        \"ollama\": \"granite3.1-dense:8b\",\n        \"hf\": \"ibm-granite/granite-3.1-8b-instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 4991620512,\n          \"minRamBytes\": 5740363589,\n          \"minVramBytes\": 5740363589,\n          \"sha256\": \"44d19d212d76a6f3fc442e8411fdb44ea6b67ceccfb00be4b4345c9a4cf813e8\"\n        }\n      ],\n      \"benchmarkProxy\": 0.47\n    },\n    {\n      \"id\": \"phi4:14b\",\n      \"family\": \"phi4\",\n      \"params\": \"14B\",\n      \"architecture\": \"dense\",\n      \"license\": \"mit\",\n      \"openWeight\": true,\n      \"contextLength\": 16384,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"reasoning\"\n      ],\n      \"releaseDate\": \"2024-12-12\",\n      \"source\": {\n        \"ollama\": \"phi4:14b\",\n        \"hf\": \"microsoft/phi-4\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 9053114464,\n          \"minRamBytes\": 10411081634,\n          \"minVramBytes\": 10411081634,\n          \"sha256\": \"fd7b6731c33c57f61767612f56517460ec2d1e2e5a3f0163e0eb3d8d8cb5df20\"\n        }\n      ],\n      \"benchmarkProxy\": 0.64\n    },\n    {\n      \"id\": \"llama3.3:70b\",\n      \"family\": \"llama3.3\",\n      \"params\": \"70B\",\n      \"architecture\": \"dense\",\n      \"license\": \"llama-3.3-community\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\",\n        \"reasoning\"\n      ],\n      \"releaseDate\": \"2024-12-06\",\n      \"source\": {\n        \"ollama\": \"llama3.3:70b\",\n        \"hf\": \"meta-llama/Llama-3.3-70B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 42520398528,\n          \"minRamBytes\": 48898458308,\n          \"minVramBytes\": 48898458308,\n          \"sha256\": \"4824460d29f2058aaf6e1118a63a7a197a09bed509f0e7d4e2efb1ee273b447d\"\n        }\n      ],\n      \"kvBytesPerToken\": 327680,\n      \"benchmarkProxy\": 0.78\n    },\n    {\n      \"id\": \"olmo2:13b\",\n      \"family\": \"olmo2\",\n      \"params\": \"13B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 4096,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2024-11-26\",\n      \"source\": {\n        \"ollama\": \"olmo2:13b\",\n        \"hf\": \"allenai/OLMo-2-1124-13B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 8354349536,\n          \"minRamBytes\": 9607501967,\n          \"minVramBytes\": 9607501967,\n          \"sha256\": \"cd836509a1a051178be134eba84115eb3a6653a1bd58473a706bf8ee4ab3a764\"\n        }\n      ],\n      \"benchmarkProxy\": 0.54\n    },\n    {\n      \"id\": \"olmo2:7b\",\n      \"family\": \"olmo2\",\n      \"params\": \"7B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 4096,\n      \"capabilities\": [\n        \"chat\"\n      ],\n      \"releaseDate\": \"2024-11-26\",\n      \"source\": {\n        \"ollama\": \"olmo2:7b\",\n        \"hf\": \"allenai/OLMo-2-1124-7B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 4472020256,\n          \"minRamBytes\": 5142823295,\n          \"minVramBytes\": 5142823295,\n          \"sha256\": \"ea89e3927d5ef671159a1359a22cdd418856c4baa2098e665f1c6eed59973968\"\n        }\n      ],\n      \"benchmarkProxy\": 0.46\n    },\n    {\n      \"id\": \"qwen2.5-coder:32b\",\n      \"family\": \"qwen2.5-coder\",\n      \"params\": \"32B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"code\",\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-11-12\",\n      \"source\": {\n        \"ollama\": \"qwen2.5-coder:32b\",\n        \"hf\": \"Qwen/Qwen2.5-Coder-32B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 19851336384,\n          \"minRamBytes\": 22829036842,\n          \"minVramBytes\": 22829036842,\n          \"sha256\": \"ac3d1ba8aa77755dab3806d9024e9c385ea0d5b412d6bdf9157f8a4a7e9fc0d9\"\n        }\n      ],\n      \"kvBytesPerToken\": 262144,\n      \"benchmarkProxy\": 0.72\n    },\n    {\n      \"id\": \"qwen2.5-coder:7b\",\n      \"family\": \"qwen2.5-coder\",\n      \"params\": \"7B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"code\",\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-11-12\",\n      \"source\": {\n        \"ollama\": \"qwen2.5-coder:7b\",\n        \"hf\": \"Qwen/Qwen2.5-Coder-7B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 4683074048,\n          \"minRamBytes\": 5385535156,\n          \"minVramBytes\": 5385535156,\n          \"sha256\": \"60e05f2100071479f596b964f89f510f057ce397ea22f2833a0cfe029bfc2463\"\n        }\n      ],\n      \"kvBytesPerToken\": 57344,\n      \"benchmarkProxy\": 0.52\n    },\n    {\n      \"id\": \"smollm2:1.7b\",\n      \"family\": \"smollm2\",\n      \"params\": \"1.7B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 8192,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-11-01\",\n      \"source\": {\n        \"ollama\": \"smollm2:1.7b\",\n        \"hf\": \"HuggingFaceTB/SmolLM2-1.7B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 1820414656,\n          \"minRamBytes\": 2093476855,\n          \"minVramBytes\": 2093476855,\n          \"sha256\": \"4d2396b16114669389d7555c15a1592aad584750310f648edad5ca8c4eccda17\"\n        }\n      ],\n      \"benchmarkProxy\": 0.2\n    },\n    {\n      \"id\": \"smollm2:135m\",\n      \"family\": \"smollm2\",\n      \"params\": \"135M\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 8192,\n      \"capabilities\": [\n        \"chat\"\n      ],\n      \"releaseDate\": \"2024-11-01\",\n      \"source\": {\n        \"ollama\": \"smollm2:135m\",\n        \"hf\": \"HuggingFaceTB/SmolLM2-135M-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 270885952,\n          \"minRamBytes\": 311518845,\n          \"minVramBytes\": 311518845,\n          \"sha256\": \"f535f83ec568d040f88ddc04a199fa6da90923bbb41d4dcaed02caa924d6ef57\"\n        }\n      ],\n      \"benchmarkProxy\": 0.05\n    },\n    {\n      \"id\": \"smollm2:360m\",\n      \"family\": \"smollm2\",\n      \"params\": \"360M\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 8192,\n      \"capabilities\": [\n        \"chat\"\n      ],\n      \"releaseDate\": \"2024-11-01\",\n      \"source\": {\n        \"ollama\": \"smollm2:360m\",\n        \"hf\": \"HuggingFaceTB/SmolLM2-360M-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 725553792,\n          \"minRamBytes\": 834386861,\n          \"minVramBytes\": 834386861,\n          \"sha256\": \"7d23be4097d67c5c43c3df62ebc19609a25f483c7d0b77d3989e1d94e36ccab6\"\n        }\n      ],\n      \"benchmarkProxy\": 0.05\n    },\n    {\n      \"id\": \"granite3-moe:3b\",\n      \"family\": \"granite3-moe\",\n      \"params\": \"3B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"0.8B\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 4096,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-10-21\",\n      \"source\": {\n        \"ollama\": \"granite3-moe:3b\",\n        \"hf\": \"ibm-granite/granite-3.0-3b-a800m-instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 2059355712,\n          \"minRamBytes\": 2368259069,\n          \"minVramBytes\": 2368259069,\n          \"sha256\": \"2c931bb3b330f26b3d4bd0ee56b828196750c244b3c40450dfc8e4074962e215\"\n        }\n      ],\n      \"benchmarkProxy\": 0.3\n    },\n    {\n      \"id\": \"llama3.2:1b\",\n      \"family\": \"llama3.2\",\n      \"params\": \"1B\",\n      \"architecture\": \"dense\",\n      \"license\": \"llama-3.2-community\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-25\",\n      \"source\": {\n        \"ollama\": \"llama3.2:1b\",\n        \"hf\": \"meta-llama/Llama-3.2-1B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 1321082688,\n          \"minRamBytes\": 1519245092,\n          \"minVramBytes\": 1519245092,\n          \"sha256\": \"74701a8c35f6c8d9a4b91f3f3497643001d63e0c7a84e085bed452548fa88d45\"\n        }\n      ],\n      \"kvBytesPerToken\": 32768,\n      \"benchmarkProxy\": 0.2\n    },\n    {\n      \"id\": \"llama3.2:3b\",\n      \"family\": \"llama3.2\",\n      \"params\": \"3B\",\n      \"architecture\": \"dense\",\n      \"license\": \"llama-3.2-community\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-25\",\n      \"source\": {\n        \"ollama\": \"llama3.2:3b\",\n        \"hf\": \"meta-llama/Llama-3.2-3B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 2019377376,\n          \"minRamBytes\": 2322283983,\n          \"minVramBytes\": 2322283983,\n          \"sha256\": \"dde5aa3fc5ffc17176b5e8bdc82f587b24b2678c6c66101bf7da77af9f7ccdff\"\n        }\n      ],\n      \"kvBytesPerToken\": 114688,\n      \"benchmarkProxy\": 0.28\n    },\n    {\n      \"id\": \"qwen2.5:0.5b\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"0.5B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"ollama\": \"qwen2.5:0.5b\",\n        \"hf\": \"Qwen/Qwen2.5-0.5B-Instruct\",\n        \"gguf\": {\n          \"repo\": \"Qwen/Qwen2.5-0.5B-Instruct-GGUF\",\n          \"revision\": \"9217f5db79a29953eb74d5343926648285ec7e67\",\n          \"file\": \"qwen2.5-0.5b-instruct-q4_k_m.gguf\",\n          \"sha256\": \"74a4da8c9fdbcd15bd1f6d01d621410d31c6fc00986f5eb687824e7b93d7a9db\"\n        }\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 397807936,\n          \"minRamBytes\": 457479127,\n          \"minVramBytes\": 457479127,\n          \"sha256\": \"c5396e06af294bd101b30dce59131a76d2b773e76950acc870eda801d3ab0515\"\n        }\n      ],\n      \"kvBytesPerToken\": 12288,\n      \"benchmarkProxy\": 0.18\n    },\n    {\n      \"id\": \"qwen2.5:0.5b-mlx\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"0.5B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"hf\": \"mlx-community/Qwen2.5-0.5B-Instruct-4bit\",\n        \"mlx\": {\n          \"repo\": \"mlx-community/Qwen2.5-0.5B-Instruct-4bit\",\n          \"revision\": \"a5339a4131f135d0fdc6a5c8b5bbed2753bbe0f3\",\n          \"files\": [\n            {\n              \"file\": \"config.json\",\n              \"sha256\": \"b045e57ea90b8f1b35f89f954b176a5c1faa02bd0af2c89bcec191239d66cef4\",\n              \"bytes\": 783\n            },\n            {\n              \"file\": \"model.safetensors\",\n              \"sha256\": \"ddffab9cbc7bf6dde941c6724841eeca8981fcfa81ca20ff8efff1396326d153\",\n              \"bytes\": 278064920\n            },\n            {\n              \"file\": \"model.safetensors.index.json\",\n              \"sha256\": \"54001cb4c11197119c206dde28e7be08e5872aab6c6d271aed339ec77e84f870\",\n              \"bytes\": 44209\n            },\n            {\n              \"file\": \"tokenizer.json\",\n              \"sha256\": \"a8506e7111b80c6d8635951a02eab0f4e1a8e4e5772da83846579e97b16f61bf\",\n              \"bytes\": 7031673\n            },\n            {\n              \"file\": \"tokenizer_config.json\",\n              \"sha256\": \"f7c61e32b7a17d19bf8e7037dcb74079a833e53ea9801f24008cac68458f03b7\",\n              \"bytes\": 7308\n            },\n            {\n              \"file\": \"special_tokens_map.json\",\n              \"sha256\": \"76862e765266b85aa9459767e33cbaf13970f327a0e88d1c65846c2ddd3a1ecd\",\n              \"bytes\": 613\n            },\n            {\n              \"file\": \"vocab.json\",\n              \"sha256\": \"ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910\",\n              \"bytes\": 2776833\n            },\n            {\n              \"file\": \"merges.txt\",\n              \"sha256\": \"8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5\",\n              \"bytes\": 1671853\n            },\n            {\n              \"file\": \"added_tokens.json\",\n              \"sha256\": \"58b54bbe36fc752f79a24a271ef66a0a0830054b4dfad94bde757d851968060b\",\n              \"bytes\": 605\n            }\n          ]\n        }\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"4bit\",\n          \"diskBytes\": 289598797,\n          \"minRamBytes\": 333038617,\n          \"minVramBytes\": 333038617\n        }\n      ],\n      \"kvBytesPerToken\": 12288,\n      \"benchmarkProxy\": 0.18\n    },\n    {\n      \"id\": \"qwen2.5:1.5b\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"1.5B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"ollama\": \"qwen2.5:1.5b\",\n        \"hf\": \"Qwen/Qwen2.5-1.5B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 986048512,\n          \"minRamBytes\": 1133955789,\n          \"minVramBytes\": 1133955789,\n          \"sha256\": \"183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4\"\n        }\n      ],\n      \"kvBytesPerToken\": 28672,\n      \"benchmarkProxy\": 0.26\n    },\n    {\n      \"id\": \"qwen2.5:14b\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"14B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"ollama\": \"qwen2.5:14b\",\n        \"hf\": \"Qwen/Qwen2.5-14B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 8988110688,\n          \"minRamBytes\": 10336327292,\n          \"minVramBytes\": 10336327292,\n          \"sha256\": \"2049f5674b1e92b4464e5729975c9689fcfbf0b0e4443ccf10b5339f370f9a54\"\n        }\n      ],\n      \"kvBytesPerToken\": 196608,\n      \"benchmarkProxy\": 0.58\n    },\n    {\n      \"id\": \"qwen2.5:32b\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"32B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"ollama\": \"qwen2.5:32b\",\n        \"hf\": \"Qwen/Qwen2.5-32B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 19851336288,\n          \"minRamBytes\": 22829036732,\n          \"minVramBytes\": 22829036732,\n          \"sha256\": \"eabc98a9bcbfce7fd70f3e07de599f8fda98120fefed5881934161ede8bd1a41\"\n        }\n      ],\n      \"kvBytesPerToken\": 262144,\n      \"benchmarkProxy\": 0.7\n    },\n    {\n      \"id\": \"qwen2.5:3b\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"3B\",\n      \"architecture\": \"dense\",\n      \"license\": \"qwen\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"ollama\": \"qwen2.5:3b\",\n        \"hf\": \"Qwen/Qwen2.5-3B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 1929903008,\n          \"minRamBytes\": 2219388460,\n          \"minVramBytes\": 2219388460,\n          \"sha256\": \"5ee4f07cdb9beadbbb293e85803c569b01bd37ed059d2715faa7bb405f31caa6\"\n        }\n      ],\n      \"kvBytesPerToken\": 36864,\n      \"benchmarkProxy\": 0.34\n    },\n    {\n      \"id\": \"qwen2.5:72b\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"72B\",\n      \"architecture\": \"dense\",\n      \"license\": \"qwen\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"ollama\": \"qwen2.5:72b\",\n        \"hf\": \"Qwen/Qwen2.5-72B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 47415715200,\n          \"minRamBytes\": 54528072480,\n          \"minVramBytes\": 54528072480,\n          \"sha256\": \"6e7fdda508e91cb0f63de5c15ff79ac63a1584ccafd751c07ca12b7f442101b8\"\n        }\n      ],\n      \"kvBytesPerToken\": 327680,\n      \"benchmarkProxy\": 0.76\n    },\n    {\n      \"id\": \"qwen2.5:7b\",\n      \"family\": \"qwen2.5\",\n      \"params\": \"7B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-09-19\",\n      \"source\": {\n        \"ollama\": \"qwen2.5:7b\",\n        \"hf\": \"Qwen/Qwen2.5-7B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 4683073952,\n          \"minRamBytes\": 5385535045,\n          \"minVramBytes\": 5385535045,\n          \"sha256\": \"2bada8a7450677000f678be90653b85d364de7db25eb5ea54136ada5f3933730\"\n        }\n      ],\n      \"kvBytesPerToken\": 57344,\n      \"benchmarkProxy\": 0.5\n    },\n    {\n      \"id\": \"phi3.5:mini\",\n      \"family\": \"phi3.5\",\n      \"params\": \"3.8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-08-20\",\n      \"source\": {\n        \"ollama\": \"phi3.5:3.8b\",\n        \"hf\": \"microsoft/Phi-3.5-mini-instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 2176177152,\n          \"minRamBytes\": 2567375000,\n          \"minVramBytes\": 2567375000,\n          \"sha256\": \"b5374915da534cb93df39f03bd4f2cd5a0c533df0d5e21957dc9556c260be9eb\"\n        }\n      ],\n      \"benchmarkProxy\": 0.36\n    },\n    {\n      \"id\": \"gemma2:2b\",\n      \"family\": \"gemma2\",\n      \"params\": \"2B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 8192,\n      \"capabilities\": [\n        \"chat\"\n      ],\n      \"releaseDate\": \"2024-07-31\",\n      \"source\": {\n        \"ollama\": \"gemma2:2b\",\n        \"hf\": \"google/gemma-2-2b-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 1629509152,\n          \"minRamBytes\": 1873935525,\n          \"minVramBytes\": 1873935525,\n          \"sha256\": \"7462734796d67c40ecec2ca98eddf970e171dbb6b370e43fd633ee75b69abe1b\"\n        }\n      ],\n      \"benchmarkProxy\": 0.32\n    },\n    {\n      \"id\": \"llama3.1:70b\",\n      \"family\": \"llama3.1\",\n      \"params\": \"70B\",\n      \"architecture\": \"dense\",\n      \"license\": \"llama-3.1-community\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-07-23\",\n      \"source\": {\n        \"ollama\": \"llama3.1:70b\",\n        \"hf\": \"meta-llama/Llama-3.1-70B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 42520398176,\n          \"minRamBytes\": 48898457903,\n          \"minVramBytes\": 48898457903,\n          \"sha256\": \"de20d2cf2dc430b1717a8b07a9df029d651f3895dbffec4729a3902a6fe344c9\"\n        }\n      ],\n      \"kvBytesPerToken\": 327680,\n      \"benchmarkProxy\": 0.74\n    },\n    {\n      \"id\": \"llama3.1:8b\",\n      \"family\": \"llama3.1\",\n      \"params\": \"8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"llama-3.1-community\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-07-23\",\n      \"source\": {\n        \"ollama\": \"llama3.1:8b\",\n        \"hf\": \"meta-llama/Llama-3.1-8B-Instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 4920738944,\n          \"minRamBytes\": 5658849786,\n          \"minVramBytes\": 5658849786,\n          \"sha256\": \"667b0c1932bc6ffc593ed1d03f895bf2dc8dc6df21db3042284a6f4416b06a29\"\n        }\n      ],\n      \"kvBytesPerToken\": 131072,\n      \"benchmarkProxy\": 0.48\n    },\n    {\n      \"id\": \"mistral-nemo:12b\",\n      \"family\": \"mistral-nemo\",\n      \"params\": \"12B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-07-18\",\n      \"source\": {\n        \"ollama\": \"mistral-nemo:12b\",\n        \"hf\": \"mistralai/Mistral-Nemo-Instruct-2407\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 7071700672,\n          \"minRamBytes\": 8132455773,\n          \"minVramBytes\": 8132455773,\n          \"sha256\": \"b559938ab7a0392fc9ea9675b82280f2a15669ec3e0e0fc491c9cb0a7681cf94\"\n        }\n      ],\n      \"kvBytesPerToken\": 163840,\n      \"benchmarkProxy\": 0.55\n    },\n    {\n      \"id\": \"gemma2:27b\",\n      \"family\": \"gemma2\",\n      \"params\": \"27B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 8192,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2024-06-27\",\n      \"source\": {\n        \"ollama\": \"gemma2:27b\",\n        \"hf\": \"google/gemma-2-27b-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 15628378336,\n          \"minRamBytes\": 18241875000,\n          \"minVramBytes\": 18241875000,\n          \"sha256\": \"d7e4b00a7d7a8d03d4eed9b0f3f61a427e9f0fc5dea6aeb414e41dee23dc8ecc\"\n        }\n      ],\n      \"benchmarkProxy\": 0.68\n    },\n    {\n      \"id\": \"gemma2:9b\",\n      \"family\": \"gemma2\",\n      \"params\": \"9B\",\n      \"architecture\": \"dense\",\n      \"license\": \"gemma\",\n      \"openWeight\": true,\n      \"contextLength\": 8192,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2024-06-27\",\n      \"source\": {\n        \"ollama\": \"gemma2:9b\",\n        \"hf\": \"google/gemma-2-9b-it\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 5443143296,\n          \"minRamBytes\": 6259614791,\n          \"minVramBytes\": 6259614791,\n          \"sha256\": \"ff1d1fc78170d787ee1201778e2dd65ea211654ca5fb7d69b5a2e7b123a50373\"\n        }\n      ],\n      \"benchmarkProxy\": 0.48\n    },\n    {\n      \"id\": \"yi:34b\",\n      \"family\": \"yi\",\n      \"params\": \"34B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 4096,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2024-05-12\",\n      \"source\": {\n        \"ollama\": \"yi:34b\",\n        \"hf\": \"01-ai/Yi-1.5-34B-Chat\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 19466529344,\n          \"minRamBytes\": 22971250000,\n          \"minVramBytes\": 22971250000,\n          \"sha256\": \"648f108f6a1ef7e12c33003d2308028ec98356d4ebda894d7db541f9e8c11aeb\"\n        }\n      ],\n      \"benchmarkProxy\": 0.67\n    },\n    {\n      \"id\": \"yi:6b\",\n      \"family\": \"yi\",\n      \"params\": \"6B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 4096,\n      \"capabilities\": [\n        \"chat\"\n      ],\n      \"releaseDate\": \"2024-05-12\",\n      \"source\": {\n        \"ollama\": \"yi:6b\",\n        \"hf\": \"01-ai/Yi-1.5-6B-Chat\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 3479326560,\n          \"minRamBytes\": 4053750000,\n          \"minVramBytes\": 4053750000,\n          \"sha256\": \"3502ee168c97b13ce3966acea684973e73c22f0ede62ca05f886b009b43b2f09\"\n        }\n      ],\n      \"benchmarkProxy\": 0.39\n    },\n    {\n      \"id\": \"yi:9b\",\n      \"family\": \"yi\",\n      \"params\": \"9B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 4096,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2024-05-12\",\n      \"source\": {\n        \"ollama\": \"yi:9b\",\n        \"hf\": \"01-ai/Yi-1.5-9B-Chat\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 5036994816,\n          \"minRamBytes\": 6080625000,\n          \"minVramBytes\": 6080625000,\n          \"sha256\": \"59b1e2a78cb0a5b8f7ba25e459ee17788242cd996e1f62a295dcbd65e5344b11\"\n        }\n      ],\n      \"benchmarkProxy\": 0.47\n    },\n    {\n      \"id\": \"phi3:mini\",\n      \"family\": \"phi3\",\n      \"params\": \"3.8B\",\n      \"architecture\": \"dense\",\n      \"license\": \"mit\",\n      \"openWeight\": true,\n      \"contextLength\": 131072,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2024-04-23\",\n      \"source\": {\n        \"ollama\": \"phi3:mini\",\n        \"hf\": \"microsoft/Phi-3-mini-128k-instruct\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 2176177120,\n          \"minRamBytes\": 2567375000,\n          \"minVramBytes\": 2567375000,\n          \"sha256\": \"633fc5be925f9a484b61d6f9b9a78021eeb462100bd557309f01ba84cac26adf\"\n        }\n      ],\n      \"benchmarkProxy\": 0.34\n    },\n    {\n      \"id\": \"mixtral:8x22b\",\n      \"family\": \"mixtral\",\n      \"params\": \"141B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"39B\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 65536,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2024-04-17\",\n      \"source\": {\n        \"ollama\": \"mixtral:8x22b\",\n        \"hf\": \"mistralai/Mixtral-8x22B-Instruct-v0.1\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 79521266688,\n          \"minRamBytes\": 95263125000,\n          \"minVramBytes\": 95263125000,\n          \"sha256\": \"94e70a590f93b8f12eb2bc6ddb2e48b3f2fab0bc577fee4b1cb42305ba590916\"\n        }\n      ],\n      \"benchmarkProxy\": 0.83\n    },\n    {\n      \"id\": \"mixtral:8x7b\",\n      \"family\": \"mixtral\",\n      \"params\": \"47B\",\n      \"architecture\": \"moe\",\n      \"activeParams\": \"13B\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\",\n        \"code\",\n        \"tools\"\n      ],\n      \"releaseDate\": \"2023-12-11\",\n      \"source\": {\n        \"ollama\": \"mixtral:8x7b\",\n        \"hf\": \"mistralai/Mixtral-8x7B-Instruct-v0.1\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 26443590560,\n          \"minRamBytes\": 31754375000,\n          \"minVramBytes\": 31754375000,\n          \"sha256\": \"f2dc41fa964b42bfe34e9fb09c0acdcfbfd6e52f1332930b4eacc9d6ad1c6cd2\"\n        }\n      ],\n      \"benchmarkProxy\": 0.75\n    },\n    {\n      \"id\": \"mistral:7b\",\n      \"family\": \"mistral\",\n      \"params\": \"7B\",\n      \"architecture\": \"dense\",\n      \"license\": \"apache-2.0\",\n      \"openWeight\": true,\n      \"contextLength\": 32768,\n      \"capabilities\": [\n        \"chat\",\n        \"code\"\n      ],\n      \"releaseDate\": \"2023-09-27\",\n      \"source\": {\n        \"ollama\": \"mistral:7b\",\n        \"hf\": \"mistralai/Mistral-7B-Instruct-v0.3\"\n      },\n      \"quantizations\": [\n        {\n          \"name\": \"Q4_K_M\",\n          \"diskBytes\": 4372811712,\n          \"minRamBytes\": 5028733469,\n          \"minVramBytes\": 5028733469,\n          \"sha256\": \"f5074b1221da0f5a2910d33b642efa5b9eb58cfdddca1c79e16d7ad28aa2b31f\"\n        }\n      ],\n      \"kvBytesPerToken\": 131072,\n      \"benchmarkProxy\": 0.46\n    }\n  ]\n}\n";
Expand description

Curated, cited model catalog bundled into every release.