[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-qwen25-72b":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":141,"fitting_gpus":213,"community":241},314,"qwen25-72b","Qwen2.5 72B","Qwen\u002FQwen2.5-72B-Instruct","Qwen","transformer",null,72700000000,"72.7B",131072,"MIT",415104,971,"published","https:\u002F\u002Fhuggingface.co\u002FQwen\u002FQwen2.5-72B-Instruct","2026-08-12T03:11:22+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",0.2614,[],[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":127},{"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",4096,135.41,176.25,264.37,250.25,"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",16384,64,304.47,456.71,315.25,{"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",8192,2,199.79,299.68,276.25,{"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},67.71,90.6,135.9,144.62,{"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},211.04,316.55,209.62,{"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},110.24,165.36,170.62,{"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},35.12,49.38,74.07,93.79,{"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},166.07,249.1,158.79,{"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},67.15,100.73,119.79,{"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},35.55,49.92,74.87,94.45,{"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},166.65,249.98,159.45,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":116,"kv_cache_gb":43,"vram_gb":84,"ram_gb":125,"disk_gb":126,"requirement_source":66,"confidence":67},101.56,120.45,{"minimum":128,"production":133,"recommended":137},{"scenario":60,"context_length":61,"batch_size":39,"weight_gb":129,"kv_cache_gb":44,"vram_gb":130,"ram_gb":131,"disk_gb":132,"requirement_source":66,"confidence":67},36.39,50.99,76.48,95.77,{"scenario":69,"context_length":70,"batch_size":33,"weight_gb":129,"kv_cache_gb":71,"vram_gb":134,"ram_gb":135,"disk_gb":136,"requirement_source":66,"confidence":67},167.82,251.73,160.77,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":129,"kv_cache_gb":43,"vram_gb":138,"ram_gb":139,"disk_gb":140,"requirement_source":66,"confidence":67},68.83,103.24,121.77,{"recommended":142,"cheapest":161,"performance":174,"min_complexity":183,"cluster_options":199,"no_match":210,"required_vram_gb":111,"candidate_count":211,"usage":212},{"gpu_model":143,"gpu_model_id":144,"gpu_count":39,"total_vram_gb":145,"hour_price":146,"monthly_cost":147,"providers":148,"bandwidth_gbps":150,"fp16_tflops":151,"interconnect_score":152,"parallel_efficiency":39,"score":153,"tier":154,"perf_metric":155,"perf_label":156,"perf_value":157,"workload":10,"cost_metric":158},"H200 SXM",11,141,3.5,2555,[149],"lambda",4800,989,40,0.5981,"balanced","tok\u002Fs","136.7 tok\u002Fs",136.7,{"effective_value":159,"effective_unit":155,"capacity_factor":160},95.7,0.7,{"gpu_model":162,"gpu_model_id":163,"gpu_count":43,"total_vram_gb":164,"hour_price":165,"monthly_cost":166,"providers":167,"bandwidth_gbps":40,"fp16_tflops":40,"interconnect_score":152,"parallel_efficiency":169,"score":170,"tier":171,"perf_metric":155,"perf_label":172,"perf_value":40,"workload":10,"cost_metric":173},"Tesla V100",17,128,0.0272,158.85,[168],"vast",0.4,0.4966,"cheapest","0 tok\u002Fs",{"effective_value":40,"effective_unit":155,"capacity_factor":160},{"gpu_model":143,"gpu_model_id":144,"gpu_count":43,"total_vram_gb":175,"hour_price":146,"monthly_cost":176,"providers":177,"bandwidth_gbps":150,"fp16_tflops":151,"interconnect_score":152,"parallel_efficiency":169,"score":178,"tier":154,"perf_metric":155,"perf_label":179,"perf_value":180,"workload":10,"cost_metric":181},1128,20440,[149],0.2197,"437.4 tok\u002Fs",437.4,{"effective_value":182,"effective_unit":155,"capacity_factor":160},306.2,{"gpu_model":184,"gpu_model_id":185,"gpu_count":39,"total_vram_gb":186,"hour_price":187,"monthly_cost":188,"providers":189,"bandwidth_gbps":192,"fp16_tflops":193,"interconnect_score":152,"parallel_efficiency":39,"score":194,"tier":154,"perf_metric":155,"perf_label":195,"perf_value":196,"workload":10,"cost_metric":197},"A100 80GB",9,80,1.2,876,[168,190,191,149],"tensordock","runpod",2039,77.97,0.6625,"58.1 tok\u002Fs",58.1,{"effective_value":198,"effective_unit":155,"capacity_factor":160},40.7,[200],{"gpu_model":201,"gpu_model_id":202,"gpu_count":43,"total_vram_gb":71,"hour_price":203,"monthly_cost":204,"providers":205,"bandwidth_gbps":206,"fp16_tflops":202,"interconnect_score":152,"parallel_efficiency":169,"insufficient":207,"vram_shortfall_gb":208,"score":209,"tier":154},"RTX 5060 Ti",24,0.0554,323.54,[168],448,true,3.15,0.5102,false,19,"value",[214,218,226,230,238],{"name":184,"vram_gb":186,"cheapest_hour":187,"cost":215,"provider":168,"estimated_tps":196},{"hourly":187,"daily":216,"monthly":188,"yearly":217},28.8,10512,{"name":219,"vram_gb":186,"cheapest_hour":220,"cost":221,"provider":168,"estimated_tps":225},"H100 SXM",1.6022,{"hourly":220,"daily":222,"monthly":223,"yearly":224},38.45,1169.61,14035.27,95.4,{"name":143,"vram_gb":145,"cheapest_hour":146,"cost":227,"provider":149,"estimated_tps":157},{"hourly":146,"daily":228,"monthly":147,"yearly":229},84,30660,{"name":231,"vram_gb":232,"cheapest_hour":233,"cost":234,"provider":168,"estimated_tps":40},"B200",180,4.1271,{"hourly":233,"daily":235,"monthly":236,"yearly":237},99.05,3012.78,36153.4,{"name":239,"vram_gb":232,"cheapest_hour":10,"cost":10,"provider":10,"estimated_tps":240},"B200 SXM",222.1,{"posts_count":40,"benchmarks_count":40}]