[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-deepseek-r1":3},{"id":4,"slug":5,"name":6,"hf_id":6,"organization":7,"architecture":8,"architecture_config":9,"parameters":14,"parameters_label":15,"context_length":16,"license":17,"downloads":18,"likes":19,"status":20,"description":21,"huggingface_url":22,"created_at":23,"seo":24,"category":26,"family":29,"score":32,"variants":33,"quantizations":34,"requirements":60,"gpu_recommendations":145,"fitting_gpus":362,"community":363},1,"deepseek-r1","deepseek-ai\u002FDeepSeek-R1","deepseek-ai","deepseek_v3",{"layers":10,"head_dim":11,"kv_heads":12,"hidden_size":13},61,56,128,7168,38537003008,"38.5B",163840,null,8044961,13571,"published","推理强化 MoE 大模型，总参数量 671B（激活 37B），思维链推理能力领先。","https:\u002F\u002Fhuggingface.co\u002Fdeepseek-ai\u002FDeepSeek-R1","2026-08-12T02:22:23+00:00",{"title":6,"description":21,"image":17,"robots":25,"canonical_url":17},"index, follow",{"name":27,"slug":28},"LLM","llm",{"name":30,"slug":31},"DeepSeek","deepseek",0.1626,[],[35,40,44,49,53,57],{"format":36,"bits":37,"size_factor":38,"quality_loss":39},"AWQ",4,0.263,0.03,{"format":41,"bits":42,"size_factor":4,"quality_loss":43},"FP16",16,0,{"format":45,"bits":46,"size_factor":47,"quality_loss":48},"FP8",8,0.5,0.01,{"format":50,"bits":37,"size_factor":51,"quality_loss":52},"GGUF",0.25,0.05,{"format":54,"bits":37,"size_factor":55,"quality_loss":56},"GPTQ",0.275,0.04,{"format":58,"bits":46,"size_factor":47,"quality_loss":59},"INT8",0.02,{"FP16":61,"FP8":85,"INT8":99,"AWQ":103,"GPTQ":117,"GGUF":131},{"minimum":62,"production":71,"recommended":78},{"scenario":63,"context_length":64,"batch_size":4,"weight_gb":65,"kv_cache_gb":47,"vram_gb":66,"ram_gb":67,"disk_gb":68,"requirement_source":69,"confidence":70},"minimum",4096,1249.83,1585.99,2378.99,1988.74,"estimated",0.95,{"scenario":72,"context_length":73,"batch_size":37,"weight_gb":65,"kv_cache_gb":74,"vram_gb":75,"ram_gb":76,"disk_gb":77,"requirement_source":69,"confidence":70},"production",16384,64,1842.37,2763.56,2053.74,{"scenario":79,"context_length":80,"batch_size":81,"weight_gb":65,"kv_cache_gb":46,"vram_gb":82,"ram_gb":83,"disk_gb":84,"requirement_source":69,"confidence":70},"recommended",8192,2,1673.61,2510.41,2014.74,{"minimum":86,"production":91,"recommended":95},{"scenario":63,"context_length":64,"batch_size":4,"weight_gb":87,"kv_cache_gb":47,"vram_gb":88,"ram_gb":89,"disk_gb":90,"requirement_source":69,"confidence":70},624.92,795.47,1193.21,1013.87,{"scenario":72,"context_length":73,"batch_size":37,"weight_gb":87,"kv_cache_gb":74,"vram_gb":92,"ram_gb":93,"disk_gb":94,"requirement_source":69,"confidence":70},979.99,1469.98,1078.87,{"scenario":79,"context_length":80,"batch_size":81,"weight_gb":87,"kv_cache_gb":46,"vram_gb":96,"ram_gb":97,"disk_gb":98,"requirement_source":69,"confidence":70},847.15,1270.73,1039.87,{"minimum":100,"production":101,"recommended":102},{"scenario":63,"context_length":64,"batch_size":4,"weight_gb":87,"kv_cache_gb":47,"vram_gb":88,"ram_gb":89,"disk_gb":90,"requirement_source":69,"confidence":70},{"scenario":72,"context_length":73,"batch_size":37,"weight_gb":87,"kv_cache_gb":74,"vram_gb":92,"ram_gb":93,"disk_gb":94,"requirement_source":69,"confidence":70},{"scenario":79,"context_length":80,"batch_size":81,"weight_gb":87,"kv_cache_gb":46,"vram_gb":96,"ram_gb":97,"disk_gb":98,"requirement_source":69,"confidence":70},{"minimum":104,"production":109,"recommended":113},{"scenario":63,"context_length":64,"batch_size":4,"weight_gb":105,"kv_cache_gb":47,"vram_gb":106,"ram_gb":107,"disk_gb":108,"requirement_source":69,"confidence":70},324.18,415.03,622.55,544.71,{"scenario":72,"context_length":73,"batch_size":37,"weight_gb":105,"kv_cache_gb":74,"vram_gb":110,"ram_gb":111,"disk_gb":112,"requirement_source":69,"confidence":70},564.96,847.44,609.71,{"scenario":79,"context_length":80,"batch_size":81,"weight_gb":105,"kv_cache_gb":46,"vram_gb":114,"ram_gb":115,"disk_gb":116,"requirement_source":69,"confidence":70},449.42,674.13,570.71,{"minimum":118,"production":123,"recommended":127},{"scenario":63,"context_length":64,"batch_size":4,"weight_gb":119,"kv_cache_gb":47,"vram_gb":120,"ram_gb":121,"disk_gb":122,"requirement_source":69,"confidence":70},328.08,419.97,629.96,550.81,{"scenario":72,"context_length":73,"batch_size":37,"weight_gb":119,"kv_cache_gb":74,"vram_gb":124,"ram_gb":125,"disk_gb":126,"requirement_source":69,"confidence":70},570.35,855.53,615.81,{"scenario":79,"context_length":80,"batch_size":81,"weight_gb":119,"kv_cache_gb":46,"vram_gb":128,"ram_gb":129,"disk_gb":130,"requirement_source":69,"confidence":70},454.59,681.88,576.81,{"minimum":132,"production":137,"recommended":141},{"scenario":63,"context_length":64,"batch_size":4,"weight_gb":133,"kv_cache_gb":47,"vram_gb":134,"ram_gb":135,"disk_gb":136,"requirement_source":69,"confidence":70},335.89,429.85,644.78,562.99,{"scenario":72,"context_length":73,"batch_size":37,"weight_gb":133,"kv_cache_gb":74,"vram_gb":138,"ram_gb":139,"disk_gb":140,"requirement_source":69,"confidence":70},581.13,871.7,627.99,{"scenario":79,"context_length":80,"batch_size":81,"weight_gb":133,"kv_cache_gb":46,"vram_gb":142,"ram_gb":143,"disk_gb":144,"requirement_source":69,"confidence":70},464.92,697.38,588.99,{"recommended":146,"cheapest":166,"performance":183,"min_complexity":193,"cluster_options":196,"no_match":360,"required_vram_gb":114,"candidate_count":37,"usage":361},{"gpu_model":147,"gpu_model_id":148,"gpu_count":46,"total_vram_gb":149,"hour_price":150,"monthly_cost":151,"providers":152,"bandwidth_gbps":154,"fp16_tflops":155,"interconnect_score":156,"parallel_efficiency":157,"score":158,"tier":159,"perf_metric":160,"perf_label":161,"perf_value":162,"workload":17,"cost_metric":163},"H200 SXM",11,1128,3.5,20440,[153],"lambda",4800,989,40,0.4,0.326,"balanced","tok\u002Fs","47.4 tok\u002Fs",47.4,{"effective_value":164,"effective_unit":160,"capacity_factor":165},33.2,0.7,{"gpu_model":167,"gpu_model_id":168,"gpu_count":46,"total_vram_gb":169,"hour_price":170,"monthly_cost":171,"providers":172,"bandwidth_gbps":176,"fp16_tflops":177,"interconnect_score":156,"parallel_efficiency":157,"score":178,"tier":159,"perf_metric":160,"perf_label":179,"perf_value":180,"workload":17,"cost_metric":181},"A100 80GB",9,640,1.2,7008,[173,174,175,153],"vast","tensordock","runpod",2039,77.97,0.4479,"20.1 tok\u002Fs",20.1,{"effective_value":182,"effective_unit":160,"capacity_factor":165},14.1,{"gpu_model":147,"gpu_model_id":148,"gpu_count":37,"total_vram_gb":184,"hour_price":150,"monthly_cost":185,"providers":186,"bandwidth_gbps":154,"fp16_tflops":155,"interconnect_score":156,"parallel_efficiency":187,"score":188,"tier":159,"perf_metric":160,"perf_label":189,"perf_value":190,"workload":17,"cost_metric":191},564,10220,[153],0.6,0.4569,"35.5 tok\u002Fs",35.5,{"effective_value":192,"effective_unit":160,"capacity_factor":165},24.8,{"gpu_model":147,"gpu_model_id":148,"gpu_count":37,"total_vram_gb":184,"hour_price":150,"monthly_cost":185,"providers":194,"bandwidth_gbps":154,"fp16_tflops":155,"interconnect_score":156,"parallel_efficiency":187,"score":188,"tier":159,"perf_metric":160,"perf_label":189,"perf_value":190,"workload":17,"cost_metric":195},[153],{"effective_value":192,"effective_unit":160,"capacity_factor":165},[197,209,218,225,232,242,253,261,269,278,287,293,302,309,316,325,334,341,351],{"gpu_model":198,"gpu_model_id":199,"gpu_count":46,"total_vram_gb":200,"hour_price":201,"monthly_cost":202,"providers":203,"bandwidth_gbps":204,"fp16_tflops":205,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":207,"score":208,"tier":159},"RTX A6000",6,384,0.2874,1678.42,[173,153],768,77,true,65.42,0.4893,{"gpu_model":210,"gpu_model_id":211,"gpu_count":46,"total_vram_gb":200,"hour_price":212,"monthly_cost":213,"providers":214,"bandwidth_gbps":215,"fp16_tflops":216,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":207,"score":217,"tier":159},"L40S",12,0.4669,2726.7,[173,174,175],864,362,0.4696,{"gpu_model":219,"gpu_model_id":220,"gpu_count":46,"total_vram_gb":200,"hour_price":221,"monthly_cost":222,"providers":223,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":207,"score":224,"tier":159},"RTX PRO 6000 WS",19,0.6689,3906.38,[173],0.4253,{"gpu_model":226,"gpu_model_id":227,"gpu_count":46,"total_vram_gb":200,"hour_price":228,"monthly_cost":229,"providers":230,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":207,"score":231,"tier":159},"RTX PRO 6000 S",20,0.8672,5064.45,[173],0.4012,{"gpu_model":233,"gpu_model_id":81,"gpu_count":46,"total_vram_gb":234,"hour_price":235,"monthly_cost":236,"providers":237,"bandwidth_gbps":238,"fp16_tflops":239,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":240,"score":241,"tier":159},"RTX 5090",256,0.3337,1948.81,[173,175],1792,209,193.42,0.507,{"gpu_model":243,"gpu_model_id":244,"gpu_count":46,"total_vram_gb":245,"hour_price":246,"monthly_cost":247,"providers":248,"bandwidth_gbps":249,"fp16_tflops":250,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":251,"score":252,"tier":159},"RTX 3090",3,192,0.0622,363.25,[173,174],936,71,257.42,0.5205,{"gpu_model":254,"gpu_model_id":4,"gpu_count":46,"total_vram_gb":245,"hour_price":255,"monthly_cost":256,"providers":257,"bandwidth_gbps":258,"fp16_tflops":259,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":251,"score":260,"tier":159},"RTX 4090",0.1356,791.9,[173,174,175,153],1008,165,0.5132,{"gpu_model":262,"gpu_model_id":263,"gpu_count":46,"total_vram_gb":245,"hour_price":264,"monthly_cost":265,"providers":266,"bandwidth_gbps":258,"fp16_tflops":267,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":251,"score":268,"tier":159},"RTX 4090D",5,0.1602,935.57,[173],82.6,0.5102,{"gpu_model":270,"gpu_model_id":271,"gpu_count":46,"total_vram_gb":245,"hour_price":272,"monthly_cost":273,"providers":274,"bandwidth_gbps":275,"fp16_tflops":276,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":251,"score":277,"tier":159},"A10",7,0.2015,1176.76,[173,175],600,125,0.4959,{"gpu_model":279,"gpu_model_id":280,"gpu_count":46,"total_vram_gb":245,"hour_price":281,"monthly_cost":282,"providers":283,"bandwidth_gbps":284,"fp16_tflops":285,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":251,"score":286,"tier":159},"L4",13,0.2022,1180.85,[173],300,121,0.4889,{"gpu_model":288,"gpu_model_id":289,"gpu_count":46,"total_vram_gb":245,"hour_price":290,"monthly_cost":291,"providers":292,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":251,"score":178,"tier":159},"RTX PRO 5000",26,0.483,2820.72,[173],{"gpu_model":294,"gpu_model_id":295,"gpu_count":46,"total_vram_gb":12,"hour_price":296,"monthly_cost":297,"providers":298,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":299,"score":300,"tier":301},"Tesla V100",17,0.0289,168.78,[173],321.42,0.5032,"cheapest",{"gpu_model":303,"gpu_model_id":304,"gpu_count":46,"total_vram_gb":12,"hour_price":305,"monthly_cost":306,"providers":307,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":299,"score":308,"tier":159},"RTX 4070S Ti",22,0.0678,395.95,[173],0.4984,{"gpu_model":310,"gpu_model_id":311,"gpu_count":46,"total_vram_gb":12,"hour_price":312,"monthly_cost":313,"providers":314,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":299,"score":315,"tier":159},"RTX 4080S",21,0.0685,400.04,[173],0.4983,{"gpu_model":317,"gpu_model_id":318,"gpu_count":46,"total_vram_gb":12,"hour_price":319,"monthly_cost":320,"providers":321,"bandwidth_gbps":322,"fp16_tflops":323,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":299,"score":324,"tier":159},"RTX 5070 Ti",18,0.0849,495.82,[173],896,88,0.5168,{"gpu_model":326,"gpu_model_id":327,"gpu_count":46,"total_vram_gb":12,"hour_price":328,"monthly_cost":329,"providers":330,"bandwidth_gbps":331,"fp16_tflops":332,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":299,"score":333,"tier":159},"RTX 5080",23,0.1339,781.98,[173],960,113,0.5123,{"gpu_model":335,"gpu_model_id":336,"gpu_count":46,"total_vram_gb":12,"hour_price":337,"monthly_cost":338,"providers":339,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":299,"score":340,"tier":159},"RTX PRO 4500",25,0.3222,1881.65,[173],0.4675,{"gpu_model":342,"gpu_model_id":343,"gpu_count":46,"total_vram_gb":344,"hour_price":345,"monthly_cost":346,"providers":347,"bandwidth_gbps":348,"fp16_tflops":10,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":349,"score":350,"tier":159},"RTX 5070",28,96,0.0804,469.54,[173],672,353.42,0.5122,{"gpu_model":352,"gpu_model_id":353,"gpu_count":46,"total_vram_gb":74,"hour_price":354,"monthly_cost":355,"providers":356,"bandwidth_gbps":357,"fp16_tflops":353,"interconnect_score":156,"parallel_efficiency":157,"insufficient":206,"vram_shortfall_gb":358,"score":359,"tier":159},"RTX 5060 Ti",24,0.0719,419.9,[173],448,385.42,0.5082,false,"value",[],{"posts_count":43,"benchmarks_count":43}]