[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-qwen-72b":3},{"id":4,"slug":5,"name":6,"hf_id":7,"organization":8,"architecture":9,"architecture_config":10,"parameters":11,"parameters_label":12,"context_length":13,"license":14,"downloads":15,"likes":16,"status":17,"description":10,"huggingface_url":18,"created_at":19,"seo":20,"category":23,"family":26,"score":28,"variants":29,"quantizations":30,"requirements":57,"gpu_recommendations":142,"fitting_gpus":214,"community":242},347,"qwen-72b","Qwen 72B","Qwen\u002FQwen-72B","Qwen","transformer",null,72000000000,"72B",32768,"MIT",2436483,360,"published","https:\u002F\u002Fhuggingface.co\u002FQwen\u002FQwen-72B","2026-08-12T03:11:24+00:00",{"title":6,"description":21,"image":10,"robots":22,"canonical_url":10},"AI 模型部署决策引擎——开源 AI 的 GPU 需求、云端价格与成本分析。","index, follow",{"name":24,"slug":25},"LLM","llm",{"name":8,"slug":27},"qwen",0.0078,[],[31,36,41,46,50,54],{"format":32,"bits":33,"size_factor":34,"quality_loss":35},"AWQ",4,0.263,0.03,{"format":37,"bits":38,"size_factor":39,"quality_loss":40},"FP16",16,1,0,{"format":42,"bits":43,"size_factor":44,"quality_loss":45},"FP8",8,0.5,0.01,{"format":47,"bits":33,"size_factor":48,"quality_loss":49},"GGUF",0.25,0.05,{"format":51,"bits":33,"size_factor":52,"quality_loss":53},"GPTQ",0.275,0.04,{"format":55,"bits":43,"size_factor":44,"quality_loss":56},"INT8",0.02,{"FP16":58,"FP8":82,"INT8":96,"AWQ":100,"GPTQ":114,"GGUF":128},{"minimum":59,"production":68,"recommended":75},{"scenario":60,"context_length":61,"batch_size":39,"weight_gb":62,"kv_cache_gb":44,"vram_gb":63,"ram_gb":64,"disk_gb":65,"requirement_source":66,"confidence":67},"minimum",4096,134.11,174.6,261.9,248.21,"estimated",0.95,{"scenario":69,"context_length":70,"batch_size":33,"weight_gb":62,"kv_cache_gb":71,"vram_gb":72,"ram_gb":73,"disk_gb":74,"requirement_source":66,"confidence":67},"production",16384,64,302.67,454.01,313.21,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":62,"kv_cache_gb":43,"vram_gb":79,"ram_gb":80,"disk_gb":81,"requirement_source":66,"confidence":67},"recommended",8192,2,198.06,297.09,274.21,{"minimum":83,"production":88,"recommended":92},{"scenario":60,"context_length":61,"batch_size":39,"weight_gb":84,"kv_cache_gb":44,"vram_gb":85,"ram_gb":86,"disk_gb":87,"requirement_source":66,"confidence":67},67.06,89.77,134.66,143.61,{"scenario":69,"context_length":70,"batch_size":33,"weight_gb":84,"kv_cache_gb":71,"vram_gb":89,"ram_gb":90,"disk_gb":91,"requirement_source":66,"confidence":67},210.14,315.2,208.61,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":84,"kv_cache_gb":43,"vram_gb":93,"ram_gb":94,"disk_gb":95,"requirement_source":66,"confidence":67},109.38,164.07,169.61,{"minimum":97,"production":98,"recommended":99},{"scenario":60,"context_length":61,"batch_size":39,"weight_gb":84,"kv_cache_gb":44,"vram_gb":85,"ram_gb":86,"disk_gb":87,"requirement_source":66,"confidence":67},{"scenario":69,"context_length":70,"batch_size":33,"weight_gb":84,"kv_cache_gb":71,"vram_gb":89,"ram_gb":90,"disk_gb":91,"requirement_source":66,"confidence":67},{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":84,"kv_cache_gb":43,"vram_gb":93,"ram_gb":94,"disk_gb":95,"requirement_source":66,"confidence":67},{"minimum":101,"production":106,"recommended":110},{"scenario":60,"context_length":61,"batch_size":39,"weight_gb":102,"kv_cache_gb":44,"vram_gb":103,"ram_gb":104,"disk_gb":105,"requirement_source":66,"confidence":67},34.78,48.95,73.43,93.26,{"scenario":69,"context_length":70,"batch_size":33,"weight_gb":102,"kv_cache_gb":71,"vram_gb":107,"ram_gb":108,"disk_gb":109,"requirement_source":66,"confidence":67},165.6,248.4,158.26,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":102,"kv_cache_gb":43,"vram_gb":111,"ram_gb":112,"disk_gb":113,"requirement_source":66,"confidence":67},66.7,100.05,119.26,{"minimum":115,"production":120,"recommended":124},{"scenario":60,"context_length":61,"batch_size":39,"weight_gb":116,"kv_cache_gb":44,"vram_gb":117,"ram_gb":118,"disk_gb":119,"requirement_source":66,"confidence":67},35.2,49.48,74.22,93.92,{"scenario":69,"context_length":70,"batch_size":33,"weight_gb":116,"kv_cache_gb":71,"vram_gb":121,"ram_gb":122,"disk_gb":123,"requirement_source":66,"confidence":67},166.18,249.27,158.92,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":116,"kv_cache_gb":43,"vram_gb":125,"ram_gb":126,"disk_gb":127,"requirement_source":66,"confidence":67},67.26,100.89,119.92,{"minimum":129,"production":134,"recommended":138},{"scenario":60,"context_length":61,"batch_size":39,"weight_gb":130,"kv_cache_gb":44,"vram_gb":131,"ram_gb":132,"disk_gb":133,"requirement_source":66,"confidence":67},36.04,50.54,75.82,95.23,{"scenario":69,"context_length":70,"batch_size":33,"weight_gb":130,"kv_cache_gb":71,"vram_gb":135,"ram_gb":136,"disk_gb":137,"requirement_source":66,"confidence":67},167.34,251.01,160.23,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":130,"kv_cache_gb":43,"vram_gb":139,"ram_gb":140,"disk_gb":141,"requirement_source":66,"confidence":67},68.37,102.55,121.23,{"recommended":143,"cheapest":162,"performance":175,"min_complexity":184,"cluster_options":200,"no_match":211,"required_vram_gb":111,"candidate_count":212,"usage":213},{"gpu_model":144,"gpu_model_id":145,"gpu_count":39,"total_vram_gb":146,"hour_price":147,"monthly_cost":148,"providers":149,"bandwidth_gbps":151,"fp16_tflops":152,"interconnect_score":153,"parallel_efficiency":39,"score":154,"tier":155,"perf_metric":156,"perf_label":157,"perf_value":158,"workload":10,"cost_metric":159},"H200 SXM",11,141,3.5,2555,[150],"lambda",4800,989,40,0.5972,"balanced","tok\u002Fs","138 tok\u002Fs",138,{"effective_value":160,"effective_unit":156,"capacity_factor":161},96.6,0.7,{"gpu_model":163,"gpu_model_id":164,"gpu_count":43,"total_vram_gb":165,"hour_price":166,"monthly_cost":167,"providers":168,"bandwidth_gbps":40,"fp16_tflops":40,"interconnect_score":153,"parallel_efficiency":170,"score":171,"tier":172,"perf_metric":156,"perf_label":173,"perf_value":40,"workload":10,"cost_metric":174},"Tesla V100",17,128,0.0272,158.85,[169],"vast",0.4,0.4956,"cheapest","0 tok\u002Fs",{"effective_value":40,"effective_unit":156,"capacity_factor":161},{"gpu_model":144,"gpu_model_id":145,"gpu_count":43,"total_vram_gb":176,"hour_price":147,"monthly_cost":177,"providers":178,"bandwidth_gbps":151,"fp16_tflops":152,"interconnect_score":153,"parallel_efficiency":170,"score":179,"tier":155,"perf_metric":156,"perf_label":180,"perf_value":181,"workload":10,"cost_metric":182},1128,20440,[150],0.2197,"441.6 tok\u002Fs",441.6,{"effective_value":183,"effective_unit":156,"capacity_factor":161},309.1,{"gpu_model":185,"gpu_model_id":186,"gpu_count":39,"total_vram_gb":187,"hour_price":188,"monthly_cost":189,"providers":190,"bandwidth_gbps":193,"fp16_tflops":194,"interconnect_score":153,"parallel_efficiency":39,"score":195,"tier":155,"perf_metric":156,"perf_label":196,"perf_value":197,"workload":10,"cost_metric":198},"A100 80GB",9,80,1.2,876,[169,191,192,150],"tensordock","runpod",2039,77.97,0.664,"58.6 tok\u002Fs",58.6,{"effective_value":199,"effective_unit":156,"capacity_factor":161},41,[201],{"gpu_model":202,"gpu_model_id":203,"gpu_count":43,"total_vram_gb":71,"hour_price":204,"monthly_cost":205,"providers":206,"bandwidth_gbps":207,"fp16_tflops":203,"interconnect_score":153,"parallel_efficiency":170,"insufficient":208,"vram_shortfall_gb":209,"score":210,"tier":155},"RTX 5060 Ti",24,0.0554,323.54,[169],448,true,2.7,0.5102,false,19,"value",[215,219,227,231,239],{"name":185,"vram_gb":187,"cheapest_hour":188,"cost":216,"provider":169,"estimated_tps":197},{"hourly":188,"daily":217,"monthly":189,"yearly":218},28.8,10512,{"name":220,"vram_gb":187,"cheapest_hour":221,"cost":222,"provider":169,"estimated_tps":226},"H100 SXM",1.6022,{"hourly":221,"daily":223,"monthly":224,"yearly":225},38.45,1169.61,14035.27,96.3,{"name":144,"vram_gb":146,"cheapest_hour":147,"cost":228,"provider":150,"estimated_tps":158},{"hourly":147,"daily":229,"monthly":148,"yearly":230},84,30660,{"name":232,"vram_gb":233,"cheapest_hour":234,"cost":235,"provider":169,"estimated_tps":40},"B200",180,4.1271,{"hourly":234,"daily":236,"monthly":237,"yearly":238},99.05,3012.78,36153.4,{"name":240,"vram_gb":233,"cheapest_hour":10,"cost":10,"provider":10,"estimated_tps":241},"B200 SXM",224.3,{"posts_count":40,"benchmarks_count":40}]