[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-qwen3-6-35b-a3b":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":17,"huggingface_url":21,"created_at":22,"seo":23,"category":26,"family":29,"score":31,"variants":32,"quantizations":33,"requirements":60,"gpu_recommendations":144,"fitting_gpus":210,"community":245},341,"qwen3-6-35b-a3b","Qwen\u002FQwen3.6-35B-A3B","Qwen","qwen3_5_moe",{"layers":10,"head_dim":11,"kv_heads":12,"hidden_size":13},40,256,2,2048,2904031232,"2.9B",262144,null,5757584,2685,"published","https:\u002F\u002Fhuggingface.co\u002FQwen\u002FQwen3.6-35B-A3B","2026-08-12T03:11:24+00:00",{"title":6,"description":24,"image":17,"robots":25,"canonical_url":17},"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":27,"slug":28},"LLM","llm",{"name":7,"slug":30},"qwen",0.0387,[],[34,39,44,49,53,57],{"format":35,"bits":36,"size_factor":37,"quality_loss":38},"AWQ",4,0.263,0.03,{"format":40,"bits":41,"size_factor":42,"quality_loss":43},"FP16",16,1,0,{"format":45,"bits":46,"size_factor":47,"quality_loss":48},"FP8",8,0.5,0.01,{"format":50,"bits":36,"size_factor":51,"quality_loss":52},"GGUF",0.25,0.05,{"format":54,"bits":36,"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":84,"INT8":98,"AWQ":102,"GPTQ":116,"GGUF":130},{"minimum":62,"production":71,"recommended":78},{"scenario":63,"context_length":64,"batch_size":42,"weight_gb":65,"kv_cache_gb":47,"vram_gb":66,"ram_gb":67,"disk_gb":68,"requirement_source":69,"confidence":70},"minimum",4096,65.19,87.42,131.13,140.7,"estimated",0.95,{"scenario":72,"context_length":73,"batch_size":36,"weight_gb":65,"kv_cache_gb":74,"vram_gb":75,"ram_gb":76,"disk_gb":77,"requirement_source":69,"confidence":70},"production",16384,64,207.57,311.35,205.7,{"scenario":79,"context_length":80,"batch_size":12,"weight_gb":65,"kv_cache_gb":46,"vram_gb":81,"ram_gb":82,"disk_gb":83,"requirement_source":69,"confidence":70},"recommended",8192,106.92,160.38,166.7,{"minimum":85,"production":90,"recommended":94},{"scenario":63,"context_length":64,"batch_size":42,"weight_gb":86,"kv_cache_gb":47,"vram_gb":87,"ram_gb":88,"disk_gb":89,"requirement_source":69,"confidence":70},32.6,46.18,69.28,89.85,{"scenario":72,"context_length":73,"batch_size":36,"weight_gb":86,"kv_cache_gb":74,"vram_gb":91,"ram_gb":92,"disk_gb":93,"requirement_source":69,"confidence":70},162.58,243.87,154.85,{"scenario":79,"context_length":80,"batch_size":12,"weight_gb":86,"kv_cache_gb":46,"vram_gb":95,"ram_gb":96,"disk_gb":97,"requirement_source":69,"confidence":70},63.81,95.71,115.85,{"minimum":99,"production":100,"recommended":101},{"scenario":63,"context_length":64,"batch_size":42,"weight_gb":86,"kv_cache_gb":47,"vram_gb":87,"ram_gb":88,"disk_gb":89,"requirement_source":69,"confidence":70},{"scenario":72,"context_length":73,"batch_size":36,"weight_gb":86,"kv_cache_gb":74,"vram_gb":91,"ram_gb":92,"disk_gb":93,"requirement_source":69,"confidence":70},{"scenario":79,"context_length":80,"batch_size":12,"weight_gb":86,"kv_cache_gb":46,"vram_gb":95,"ram_gb":96,"disk_gb":97,"requirement_source":69,"confidence":70},{"minimum":103,"production":108,"recommended":112},{"scenario":63,"context_length":64,"batch_size":42,"weight_gb":104,"kv_cache_gb":47,"vram_gb":105,"ram_gb":106,"disk_gb":107,"requirement_source":69,"confidence":70},16.91,26.34,39.51,65.38,{"scenario":72,"context_length":73,"batch_size":36,"weight_gb":104,"kv_cache_gb":74,"vram_gb":109,"ram_gb":110,"disk_gb":111,"requirement_source":69,"confidence":70},140.93,211.4,130.38,{"scenario":79,"context_length":80,"batch_size":12,"weight_gb":104,"kv_cache_gb":46,"vram_gb":113,"ram_gb":114,"disk_gb":115,"requirement_source":69,"confidence":70},43.06,64.59,91.38,{"minimum":117,"production":122,"recommended":126},{"scenario":63,"context_length":64,"batch_size":42,"weight_gb":118,"kv_cache_gb":47,"vram_gb":119,"ram_gb":120,"disk_gb":121,"requirement_source":69,"confidence":70},17.11,26.6,39.9,65.7,{"scenario":72,"context_length":73,"batch_size":36,"weight_gb":118,"kv_cache_gb":74,"vram_gb":123,"ram_gb":124,"disk_gb":125,"requirement_source":69,"confidence":70},141.22,211.82,130.7,{"scenario":79,"context_length":80,"batch_size":12,"weight_gb":118,"kv_cache_gb":46,"vram_gb":127,"ram_gb":128,"disk_gb":129,"requirement_source":69,"confidence":70},43.33,65,91.7,{"minimum":131,"production":136,"recommended":140},{"scenario":63,"context_length":64,"batch_size":42,"weight_gb":132,"kv_cache_gb":47,"vram_gb":133,"ram_gb":134,"disk_gb":135,"requirement_source":69,"confidence":70},17.52,27.11,40.67,66.33,{"scenario":72,"context_length":73,"batch_size":36,"weight_gb":132,"kv_cache_gb":74,"vram_gb":137,"ram_gb":138,"disk_gb":139,"requirement_source":69,"confidence":70},141.78,212.67,131.33,{"scenario":79,"context_length":80,"batch_size":12,"weight_gb":132,"kv_cache_gb":46,"vram_gb":141,"ram_gb":142,"disk_gb":143,"requirement_source":69,"confidence":70},43.87,65.81,92.33,{"recommended":145,"cheapest":166,"performance":177,"min_complexity":192,"cluster_options":206,"no_match":207,"required_vram_gb":113,"candidate_count":208,"usage":209},{"gpu_model":146,"gpu_model_id":147,"gpu_count":42,"total_vram_gb":148,"hour_price":149,"monthly_cost":150,"providers":151,"bandwidth_gbps":156,"fp16_tflops":157,"interconnect_score":10,"parallel_efficiency":42,"score":158,"tier":159,"perf_metric":160,"perf_label":161,"perf_value":162,"workload":17,"cost_metric":163},"H100 SXM",10,80,1.6022,1169.61,[152,153,154,155],"vast","tensordock","lambda","runpod",3350,66.9,0.6331,"balanced","tok\u002Fs","198.1 tok\u002Fs",198.1,{"effective_value":164,"effective_unit":160,"capacity_factor":165},138.7,0.7,{"gpu_model":167,"gpu_model_id":168,"gpu_count":36,"total_vram_gb":74,"hour_price":169,"monthly_cost":170,"providers":171,"bandwidth_gbps":43,"fp16_tflops":43,"interconnect_score":10,"parallel_efficiency":172,"score":173,"tier":174,"perf_metric":160,"perf_label":175,"perf_value":43,"workload":17,"cost_metric":176},"Tesla V100",17,0.0272,79.42,[152],0.6,0.5778,"cheapest","0 tok\u002Fs",{"effective_value":43,"effective_unit":160,"capacity_factor":165},{"gpu_model":178,"gpu_model_id":179,"gpu_count":46,"total_vram_gb":180,"hour_price":181,"monthly_cost":182,"providers":183,"bandwidth_gbps":184,"fp16_tflops":185,"interconnect_score":10,"parallel_efficiency":186,"score":187,"tier":159,"perf_metric":160,"perf_label":188,"perf_value":189,"workload":17,"cost_metric":190},"H200 SXM",11,1128,3.5,20440,[154],4800,989,0.4,0.2197,"908.3 tok\u002Fs",908.3,{"effective_value":191,"effective_unit":160,"capacity_factor":165},635.8,{"gpu_model":193,"gpu_model_id":194,"gpu_count":42,"total_vram_gb":195,"hour_price":196,"monthly_cost":197,"providers":198,"bandwidth_gbps":199,"fp16_tflops":200,"interconnect_score":10,"parallel_efficiency":42,"score":201,"tier":174,"perf_metric":160,"perf_label":202,"perf_value":203,"workload":17,"cost_metric":204},"RTX A6000",6,48,0.3001,219.07,[152,154],768,77,0.6517,"45.4 tok\u002Fs",45.4,{"effective_value":205,"effective_unit":160,"capacity_factor":165},31.8,[],false,23,"value",[211,215,223,230,237],{"name":193,"vram_gb":195,"cheapest_hour":196,"cost":212,"provider":152,"estimated_tps":203},{"hourly":196,"daily":213,"monthly":197,"yearly":214},7.2,2628.88,{"name":216,"vram_gb":195,"cheapest_hour":217,"cost":218,"provider":152,"estimated_tps":222},"L40S",0.4669,{"hourly":217,"daily":219,"monthly":220,"yearly":221},11.21,340.84,4090.04,51.1,{"name":224,"vram_gb":195,"cheapest_hour":225,"cost":226,"provider":152,"estimated_tps":43},"RTX PRO 6000 WS",0.8022,{"hourly":225,"daily":227,"monthly":228,"yearly":229},19.25,585.61,7027.27,{"name":231,"vram_gb":195,"cheapest_hour":232,"cost":233,"provider":152,"estimated_tps":43},"RTX PRO 6000 S",1.0131,{"hourly":232,"daily":234,"monthly":235,"yearly":236},24.31,739.56,8874.76,{"name":238,"vram_gb":148,"cheapest_hour":239,"cost":240,"provider":152,"estimated_tps":244},"A100 80GB",1.2,{"hourly":239,"daily":241,"monthly":242,"yearly":243},28.8,876,10512,120.6,{"posts_count":43,"benchmarks_count":43}]