[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-starcoder2-15b":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":10,"score":26,"variants":27,"quantizations":28,"requirements":55,"gpu_recommendations":137,"fitting_gpus":204,"community":239},372,"starcoder2-15b","StarCoder2 15B","bigcode\u002Fstarcoder2-15b","bigcode","transformer",null,15000000000,"15B",16384,"MIT",3827,674,"published","https:\u002F\u002Fhuggingface.co\u002Fbigcode\u002Fstarcoder2-15b","2026-08-12T03:11:25+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",0.0075,[],[29,34,39,44,48,52],{"format":30,"bits":31,"size_factor":32,"quality_loss":33},"AWQ",4,0.263,0.03,{"format":35,"bits":36,"size_factor":37,"quality_loss":38},"FP16",16,1,0,{"format":40,"bits":41,"size_factor":42,"quality_loss":43},"FP8",8,0.5,0.01,{"format":45,"bits":31,"size_factor":46,"quality_loss":47},"GGUF",0.25,0.05,{"format":49,"bits":31,"size_factor":50,"quality_loss":51},"GPTQ",0.275,0.04,{"format":53,"bits":41,"size_factor":42,"quality_loss":54},"INT8",0.02,{"FP16":56,"FP8":79,"INT8":93,"AWQ":97,"GPTQ":111,"GGUF":124},{"minimum":57,"production":66,"recommended":72},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":60,"kv_cache_gb":42,"vram_gb":61,"ram_gb":62,"disk_gb":63,"requirement_source":64,"confidence":65},"minimum",4096,27.94,40.29,60.44,82.59,"estimated",0.95,{"scenario":67,"context_length":13,"batch_size":31,"weight_gb":60,"kv_cache_gb":68,"vram_gb":69,"ram_gb":70,"disk_gb":71,"requirement_source":64,"confidence":65},"production",64,156.16,234.24,147.59,{"scenario":73,"context_length":74,"batch_size":75,"weight_gb":60,"kv_cache_gb":41,"vram_gb":76,"ram_gb":77,"disk_gb":78,"requirement_source":64,"confidence":65},"recommended",8192,2,57.65,86.48,108.59,{"minimum":80,"production":85,"recommended":89},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":81,"kv_cache_gb":42,"vram_gb":82,"ram_gb":83,"disk_gb":84,"requirement_source":64,"confidence":65},13.97,22.62,33.93,60.79,{"scenario":67,"context_length":13,"batch_size":31,"weight_gb":81,"kv_cache_gb":68,"vram_gb":86,"ram_gb":87,"disk_gb":88,"requirement_source":64,"confidence":65},136.88,205.32,125.79,{"scenario":73,"context_length":74,"batch_size":75,"weight_gb":81,"kv_cache_gb":41,"vram_gb":90,"ram_gb":91,"disk_gb":92,"requirement_source":64,"confidence":65},39.18,58.76,86.79,{"minimum":94,"production":95,"recommended":96},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":81,"kv_cache_gb":42,"vram_gb":82,"ram_gb":83,"disk_gb":84,"requirement_source":64,"confidence":65},{"scenario":67,"context_length":13,"batch_size":31,"weight_gb":81,"kv_cache_gb":68,"vram_gb":86,"ram_gb":87,"disk_gb":88,"requirement_source":64,"confidence":65},{"scenario":73,"context_length":74,"batch_size":75,"weight_gb":81,"kv_cache_gb":41,"vram_gb":90,"ram_gb":91,"disk_gb":92,"requirement_source":64,"confidence":65},{"minimum":98,"production":103,"recommended":107},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":99,"kv_cache_gb":42,"vram_gb":100,"ram_gb":101,"disk_gb":102,"requirement_source":64,"confidence":65},7.25,14.12,32,50.31,{"scenario":67,"context_length":13,"batch_size":31,"weight_gb":99,"kv_cache_gb":68,"vram_gb":104,"ram_gb":105,"disk_gb":106,"requirement_source":64,"confidence":65},127.6,191.4,115.31,{"scenario":73,"context_length":74,"batch_size":75,"weight_gb":99,"kv_cache_gb":41,"vram_gb":108,"ram_gb":109,"disk_gb":110,"requirement_source":64,"confidence":65},30.28,45.43,76.31,{"minimum":112,"production":116,"recommended":120},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":113,"kv_cache_gb":42,"vram_gb":114,"ram_gb":101,"disk_gb":115,"requirement_source":64,"confidence":65},7.33,14.23,50.44,{"scenario":67,"context_length":13,"batch_size":31,"weight_gb":113,"kv_cache_gb":68,"vram_gb":117,"ram_gb":118,"disk_gb":119,"requirement_source":64,"confidence":65},127.72,191.58,115.44,{"scenario":73,"context_length":74,"batch_size":75,"weight_gb":113,"kv_cache_gb":41,"vram_gb":121,"ram_gb":122,"disk_gb":123,"requirement_source":64,"confidence":65},30.4,45.6,76.44,{"minimum":125,"production":129,"recommended":133},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":126,"kv_cache_gb":42,"vram_gb":127,"ram_gb":101,"disk_gb":128,"requirement_source":64,"confidence":65},7.51,14.45,50.71,{"scenario":67,"context_length":13,"batch_size":31,"weight_gb":126,"kv_cache_gb":68,"vram_gb":130,"ram_gb":131,"disk_gb":132,"requirement_source":64,"confidence":65},127.96,191.94,115.71,{"scenario":73,"context_length":74,"batch_size":75,"weight_gb":126,"kv_cache_gb":41,"vram_gb":134,"ram_gb":135,"disk_gb":136,"requirement_source":64,"confidence":65},30.63,45.95,76.71,{"recommended":138,"cheapest":160,"performance":171,"min_complexity":186,"cluster_options":200,"no_match":201,"required_vram_gb":108,"candidate_count":202,"usage":203},{"gpu_model":139,"gpu_model_id":140,"gpu_count":37,"total_vram_gb":141,"hour_price":142,"monthly_cost":143,"providers":144,"bandwidth_gbps":149,"fp16_tflops":150,"interconnect_score":151,"parallel_efficiency":37,"score":152,"tier":153,"perf_metric":154,"perf_label":155,"perf_value":156,"workload":10,"cost_metric":157},"H100 SXM",10,80,1.6022,1169.61,[145,146,147,148],"vast","tensordock","lambda","runpod",3350,66.9,40,0.5905,"balanced","tok\u002Fs","462.1 tok\u002Fs",462.1,{"effective_value":158,"effective_unit":154,"capacity_factor":159},323.5,0.7,{"gpu_model":161,"gpu_model_id":162,"gpu_count":75,"total_vram_gb":101,"hour_price":163,"monthly_cost":164,"providers":165,"bandwidth_gbps":38,"fp16_tflops":38,"interconnect_score":151,"parallel_efficiency":166,"score":167,"tier":168,"perf_metric":154,"perf_label":169,"perf_value":38,"workload":10,"cost_metric":170},"Tesla V100",17,0.0272,39.71,[145],0.8,0.5943,"cheapest","0 tok\u002Fs",{"effective_value":38,"effective_unit":154,"capacity_factor":159},{"gpu_model":172,"gpu_model_id":173,"gpu_count":41,"total_vram_gb":174,"hour_price":175,"monthly_cost":176,"providers":177,"bandwidth_gbps":178,"fp16_tflops":179,"interconnect_score":151,"parallel_efficiency":180,"score":181,"tier":153,"perf_metric":154,"perf_label":182,"perf_value":183,"workload":10,"cost_metric":184},"H200 SXM",11,1128,3.5,20440,[147],4800,989,0.4,0.2197,"2118.6 tok\u002Fs",2118.6,{"effective_value":185,"effective_unit":154,"capacity_factor":159},1483,{"gpu_model":187,"gpu_model_id":188,"gpu_count":37,"total_vram_gb":189,"hour_price":190,"monthly_cost":191,"providers":192,"bandwidth_gbps":193,"fp16_tflops":194,"interconnect_score":151,"parallel_efficiency":37,"score":195,"tier":168,"perf_metric":154,"perf_label":196,"perf_value":197,"workload":10,"cost_metric":198},"RTX A6000",6,48,0.3001,219.07,[145,147],768,77,0.6591,"105.9 tok\u002Fs",105.9,{"effective_value":199,"effective_unit":154,"capacity_factor":159},74.1,[],false,24,"value",[205,209,217,225,232],{"name":187,"vram_gb":189,"cheapest_hour":190,"cost":206,"provider":145,"estimated_tps":197},{"hourly":190,"daily":207,"monthly":191,"yearly":208},7.2,2628.88,{"name":210,"vram_gb":101,"cheapest_hour":211,"cost":212,"provider":145,"estimated_tps":216},"RTX 5090",0.3222,{"hourly":211,"daily":213,"monthly":214,"yearly":215},7.73,235.21,2822.47,247.2,{"name":218,"vram_gb":189,"cheapest_hour":219,"cost":220,"provider":145,"estimated_tps":224},"L40S",0.4669,{"hourly":219,"daily":221,"monthly":222,"yearly":223},11.21,340.84,4090.04,119.2,{"name":226,"vram_gb":189,"cheapest_hour":227,"cost":228,"provider":145,"estimated_tps":38},"RTX PRO 6000 WS",0.8022,{"hourly":227,"daily":229,"monthly":230,"yearly":231},19.25,585.61,7027.27,{"name":233,"vram_gb":189,"cheapest_hour":234,"cost":235,"provider":145,"estimated_tps":38},"RTX PRO 6000 S",1.0131,{"hourly":234,"daily":236,"monthly":237,"yearly":238},24.31,739.56,8874.76,{"posts_count":38,"benchmarks_count":38}]