[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-deepseek-coder-67b":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":29,"variants":30,"quantizations":31,"requirements":58,"gpu_recommendations":138,"fitting_gpus":202,"community":233},89,"deepseek-coder-67b","DeepSeek-Coder 6.7B","deepseek-ai\u002Fdeepseek-coder-6.7b-instruct","deepseek-ai","transformer",null,6700000000,"6.7B",16384,"MIT",510214,508,"published","https:\u002F\u002Fhuggingface.co\u002Fdeepseek-ai\u002Fdeepseek-coder-6.7b-instruct","2026-08-12T03:11:10+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},"Coding AI","coding",{"name":27,"slug":28},"DeepSeek","deepseek",0.0275,[],[32,37,42,47,51,55],{"format":33,"bits":34,"size_factor":35,"quality_loss":36},"AWQ",4,0.263,0.03,{"format":38,"bits":39,"size_factor":40,"quality_loss":41},"FP16",16,1,0,{"format":43,"bits":44,"size_factor":45,"quality_loss":46},"FP8",8,0.5,0.01,{"format":48,"bits":34,"size_factor":49,"quality_loss":50},"GGUF",0.25,0.05,{"format":52,"bits":34,"size_factor":53,"quality_loss":54},"GPTQ",0.275,0.04,{"format":56,"bits":44,"size_factor":45,"quality_loss":57},"INT8",0.02,{"FP16":59,"FP8":82,"INT8":95,"AWQ":99,"GPTQ":112,"GGUF":125},{"minimum":60,"production":69,"recommended":75},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":63,"kv_cache_gb":45,"vram_gb":64,"ram_gb":65,"disk_gb":66,"requirement_source":67,"confidence":68},"minimum",4096,12.48,20.74,32,58.47,"estimated",0.95,{"scenario":70,"context_length":13,"batch_size":34,"weight_gb":63,"kv_cache_gb":71,"vram_gb":72,"ram_gb":73,"disk_gb":74,"requirement_source":67,"confidence":68},"production",64,134.82,202.23,123.47,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":63,"kv_cache_gb":44,"vram_gb":79,"ram_gb":80,"disk_gb":81,"requirement_source":67,"confidence":68},"recommended",8192,2,37.2,55.81,84.47,{"minimum":83,"production":87,"recommended":91},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":84,"kv_cache_gb":45,"vram_gb":85,"ram_gb":65,"disk_gb":86,"requirement_source":67,"confidence":68},6.24,12.84,48.73,{"scenario":70,"context_length":13,"batch_size":34,"weight_gb":84,"kv_cache_gb":71,"vram_gb":88,"ram_gb":89,"disk_gb":90,"requirement_source":67,"confidence":68},126.21,189.32,113.73,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":84,"kv_cache_gb":44,"vram_gb":92,"ram_gb":93,"disk_gb":94,"requirement_source":67,"confidence":68},28.95,43.43,74.73,{"minimum":96,"production":97,"recommended":98},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":84,"kv_cache_gb":45,"vram_gb":85,"ram_gb":65,"disk_gb":86,"requirement_source":67,"confidence":68},{"scenario":70,"context_length":13,"batch_size":34,"weight_gb":84,"kv_cache_gb":71,"vram_gb":88,"ram_gb":89,"disk_gb":90,"requirement_source":67,"confidence":68},{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":84,"kv_cache_gb":44,"vram_gb":92,"ram_gb":93,"disk_gb":94,"requirement_source":67,"confidence":68},{"minimum":100,"production":104,"recommended":108},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":101,"kv_cache_gb":45,"vram_gb":102,"ram_gb":65,"disk_gb":103,"requirement_source":67,"confidence":68},3.24,9.04,44.05,{"scenario":70,"context_length":13,"batch_size":34,"weight_gb":101,"kv_cache_gb":71,"vram_gb":105,"ram_gb":106,"disk_gb":107,"requirement_source":67,"confidence":68},122.07,183.1,109.05,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":101,"kv_cache_gb":44,"vram_gb":109,"ram_gb":110,"disk_gb":111,"requirement_source":67,"confidence":68},24.98,37.47,70.05,{"minimum":113,"production":117,"recommended":121},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":114,"kv_cache_gb":45,"vram_gb":115,"ram_gb":65,"disk_gb":116,"requirement_source":67,"confidence":68},3.28,9.09,44.11,{"scenario":70,"context_length":13,"batch_size":34,"weight_gb":114,"kv_cache_gb":71,"vram_gb":118,"ram_gb":119,"disk_gb":120,"requirement_source":67,"confidence":68},122.12,183.18,109.11,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":114,"kv_cache_gb":44,"vram_gb":122,"ram_gb":123,"disk_gb":124,"requirement_source":67,"confidence":68},25.03,37.55,70.11,{"minimum":126,"production":130,"recommended":134},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":127,"kv_cache_gb":45,"vram_gb":128,"ram_gb":65,"disk_gb":129,"requirement_source":67,"confidence":68},3.35,9.19,44.23,{"scenario":70,"context_length":13,"batch_size":34,"weight_gb":127,"kv_cache_gb":71,"vram_gb":131,"ram_gb":132,"disk_gb":133,"requirement_source":67,"confidence":68},122.23,183.34,109.23,{"scenario":76,"context_length":77,"batch_size":78,"weight_gb":127,"kv_cache_gb":44,"vram_gb":135,"ram_gb":136,"disk_gb":137,"requirement_source":67,"confidence":68},25.14,37.7,70.23,{"recommended":139,"cheapest":157,"performance":167,"min_complexity":184,"cluster_options":198,"no_match":199,"required_vram_gb":109,"candidate_count":200,"usage":201},{"gpu_model":140,"gpu_model_id":78,"gpu_count":40,"total_vram_gb":65,"hour_price":141,"monthly_cost":142,"providers":143,"bandwidth_gbps":146,"fp16_tflops":147,"interconnect_score":148,"parallel_efficiency":40,"score":149,"tier":150,"perf_metric":151,"perf_label":152,"perf_value":153,"workload":10,"cost_metric":154},"RTX 5090",0.3222,235.21,[144,145],"vast","runpod",1792,209,40,0.6897,"cheapest","tok\u002Fs","553.1 tok\u002Fs",553.1,{"effective_value":155,"effective_unit":151,"capacity_factor":156},387.2,0.7,{"gpu_model":158,"gpu_model_id":159,"gpu_count":78,"total_vram_gb":65,"hour_price":160,"monthly_cost":161,"providers":162,"bandwidth_gbps":41,"fp16_tflops":41,"interconnect_score":148,"parallel_efficiency":163,"score":164,"tier":150,"perf_metric":151,"perf_label":165,"perf_value":41,"workload":10,"cost_metric":166},"Tesla V100",17,0.0272,39.71,[144],0.8,0.6385,"0 tok\u002Fs",{"effective_value":41,"effective_unit":151,"capacity_factor":156},{"gpu_model":168,"gpu_model_id":169,"gpu_count":44,"total_vram_gb":170,"hour_price":171,"monthly_cost":172,"providers":173,"bandwidth_gbps":175,"fp16_tflops":176,"interconnect_score":148,"parallel_efficiency":177,"score":178,"tier":179,"perf_metric":151,"perf_label":180,"perf_value":181,"workload":10,"cost_metric":182},"H200 SXM",11,1128,3.5,20440,[174],"lambda",4800,989,0.4,0.2197,"balanced","4740.7 tok\u002Fs",4740.7,{"effective_value":183,"effective_unit":151,"capacity_factor":156},3318.5,{"gpu_model":185,"gpu_model_id":186,"gpu_count":40,"total_vram_gb":187,"hour_price":188,"monthly_cost":189,"providers":190,"bandwidth_gbps":191,"fp16_tflops":192,"interconnect_score":148,"parallel_efficiency":40,"score":193,"tier":150,"perf_metric":151,"perf_label":194,"perf_value":195,"workload":10,"cost_metric":196},"RTX A6000",6,48,0.3001,219.07,[144,174],768,77,0.6297,"237 tok\u002Fs",237,{"effective_value":197,"effective_unit":151,"capacity_factor":156},165.9,[],false,24,"value",[203,207,211,219,226],{"name":185,"vram_gb":187,"cheapest_hour":188,"cost":204,"provider":144,"estimated_tps":195},{"hourly":188,"daily":205,"monthly":189,"yearly":206},7.2,2628.88,{"name":140,"vram_gb":65,"cheapest_hour":141,"cost":208,"provider":144,"estimated_tps":153},{"hourly":141,"daily":209,"monthly":142,"yearly":210},7.73,2822.47,{"name":212,"vram_gb":187,"cheapest_hour":213,"cost":214,"provider":144,"estimated_tps":218},"L40S",0.4669,{"hourly":213,"daily":215,"monthly":216,"yearly":217},11.21,340.84,4090.04,266.7,{"name":220,"vram_gb":187,"cheapest_hour":221,"cost":222,"provider":144,"estimated_tps":41},"RTX PRO 6000 WS",0.8022,{"hourly":221,"daily":223,"monthly":224,"yearly":225},19.25,585.61,7027.27,{"name":227,"vram_gb":187,"cheapest_hour":228,"cost":229,"provider":144,"estimated_tps":41},"RTX PRO 6000 S",1.0131,{"hourly":228,"daily":230,"monthly":231,"yearly":232},24.31,739.56,8874.76,{"posts_count":41,"benchmarks_count":41}]