[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-exaone-35-78b":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":201,"community":232},115,"exaone-35-78b","EXAONE 3.5 7.8B","LGAI-EXAONE\u002FEXAONE-3.5-7.8B-Instruct","LGAI-EXAONE","transformer",null,7800000000,"7.8B",131072,"MIT",519146,159,"published","https:\u002F\u002Fhuggingface.co\u002FLGAI-EXAONE\u002FEXAONE-3.5-7.8B-Instruct","2026-08-12T03:11:12+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",0.2526,[],[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":80,"INT8":94,"AWQ":98,"GPTQ":111,"GGUF":124},{"minimum":57,"production":66,"recommended":73},{"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",4096,14.53,23.33,34.99,61.66,"estimated",0.95,{"scenario":67,"context_length":68,"batch_size":31,"weight_gb":60,"kv_cache_gb":69,"vram_gb":70,"ram_gb":71,"disk_gb":72,"requirement_source":64,"confidence":65},"production",16384,64,137.65,206.47,126.66,{"scenario":74,"context_length":75,"batch_size":76,"weight_gb":60,"kv_cache_gb":41,"vram_gb":77,"ram_gb":78,"disk_gb":79,"requirement_source":64,"confidence":65},"recommended",8192,2,39.91,59.87,87.66,{"minimum":81,"production":86,"recommended":90},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":82,"kv_cache_gb":42,"vram_gb":83,"ram_gb":84,"disk_gb":85,"requirement_source":64,"confidence":65},7.26,14.14,32,50.33,{"scenario":67,"context_length":68,"batch_size":31,"weight_gb":82,"kv_cache_gb":69,"vram_gb":87,"ram_gb":88,"disk_gb":89,"requirement_source":64,"confidence":65},127.62,191.44,115.33,{"scenario":74,"context_length":75,"batch_size":76,"weight_gb":82,"kv_cache_gb":41,"vram_gb":91,"ram_gb":92,"disk_gb":93,"requirement_source":64,"confidence":65},30.31,45.46,76.33,{"minimum":95,"production":96,"recommended":97},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":82,"kv_cache_gb":42,"vram_gb":83,"ram_gb":84,"disk_gb":85,"requirement_source":64,"confidence":65},{"scenario":67,"context_length":68,"batch_size":31,"weight_gb":82,"kv_cache_gb":69,"vram_gb":87,"ram_gb":88,"disk_gb":89,"requirement_source":64,"confidence":65},{"scenario":74,"context_length":75,"batch_size":76,"weight_gb":82,"kv_cache_gb":41,"vram_gb":91,"ram_gb":92,"disk_gb":93,"requirement_source":64,"confidence":65},{"minimum":99,"production":103,"recommended":107},{"scenario":58,"context_length":59,"batch_size":37,"weight_gb":100,"kv_cache_gb":42,"vram_gb":101,"ram_gb":84,"disk_gb":102,"requirement_source":64,"confidence":65},3.77,9.72,44.88,{"scenario":67,"context_length":68,"batch_size":31,"weight_gb":100,"kv_cache_gb":69,"vram_gb":104,"ram_gb":105,"disk_gb":106,"requirement_source":64,"confidence":65},122.8,184.2,109.88,{"scenario":74,"context_length":75,"batch_size":76,"weight_gb":100,"kv_cache_gb":41,"vram_gb":108,"ram_gb":109,"disk_gb":110,"requirement_source":64,"confidence":65},25.68,38.53,70.88,{"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":84,"disk_gb":115,"requirement_source":64,"confidence":65},3.81,9.77,44.95,{"scenario":67,"context_length":68,"batch_size":31,"weight_gb":113,"kv_cache_gb":69,"vram_gb":117,"ram_gb":118,"disk_gb":119,"requirement_source":64,"confidence":65},122.86,184.29,109.95,{"scenario":74,"context_length":75,"batch_size":76,"weight_gb":113,"kv_cache_gb":41,"vram_gb":121,"ram_gb":122,"disk_gb":123,"requirement_source":64,"confidence":65},25.74,38.62,70.95,{"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":84,"disk_gb":128,"requirement_source":64,"confidence":65},3.9,9.89,45.09,{"scenario":67,"context_length":68,"batch_size":31,"weight_gb":126,"kv_cache_gb":69,"vram_gb":130,"ram_gb":131,"disk_gb":132,"requirement_source":64,"confidence":65},122.99,184.48,110.09,{"scenario":74,"context_length":75,"batch_size":76,"weight_gb":126,"kv_cache_gb":41,"vram_gb":134,"ram_gb":135,"disk_gb":136,"requirement_source":64,"confidence":65},25.86,38.8,71.09,{"recommended":138,"cheapest":156,"performance":166,"min_complexity":183,"cluster_options":197,"no_match":198,"required_vram_gb":108,"candidate_count":199,"usage":200},{"gpu_model":139,"gpu_model_id":76,"gpu_count":37,"total_vram_gb":84,"hour_price":140,"monthly_cost":141,"providers":142,"bandwidth_gbps":145,"fp16_tflops":146,"interconnect_score":147,"parallel_efficiency":37,"score":148,"tier":149,"perf_metric":150,"perf_label":151,"perf_value":152,"workload":10,"cost_metric":153},"RTX 5090",0.3222,235.21,[143,144],"vast","runpod",1792,209,40,0.6839,"cheapest","tok\u002Fs","475.3 tok\u002Fs",475.3,{"effective_value":154,"effective_unit":150,"capacity_factor":155},332.7,0.7,{"gpu_model":157,"gpu_model_id":158,"gpu_count":76,"total_vram_gb":84,"hour_price":159,"monthly_cost":160,"providers":161,"bandwidth_gbps":38,"fp16_tflops":38,"interconnect_score":147,"parallel_efficiency":162,"score":163,"tier":149,"perf_metric":150,"perf_label":164,"perf_value":38,"workload":10,"cost_metric":165},"Tesla V100",17,0.0272,39.71,[143],0.8,0.6327,"0 tok\u002Fs",{"effective_value":38,"effective_unit":150,"capacity_factor":155},{"gpu_model":167,"gpu_model_id":168,"gpu_count":41,"total_vram_gb":169,"hour_price":170,"monthly_cost":171,"providers":172,"bandwidth_gbps":174,"fp16_tflops":175,"interconnect_score":147,"parallel_efficiency":176,"score":177,"tier":178,"perf_metric":150,"perf_label":179,"perf_value":180,"workload":10,"cost_metric":181},"H200 SXM",11,1128,3.5,20440,[173],"lambda",4800,989,0.4,0.2197,"balanced","4074.3 tok\u002Fs",4074.3,{"effective_value":182,"effective_unit":150,"capacity_factor":155},2852,{"gpu_model":184,"gpu_model_id":185,"gpu_count":37,"total_vram_gb":186,"hour_price":187,"monthly_cost":188,"providers":189,"bandwidth_gbps":190,"fp16_tflops":191,"interconnect_score":147,"parallel_efficiency":37,"score":192,"tier":149,"perf_metric":150,"perf_label":193,"perf_value":194,"workload":10,"cost_metric":195},"RTX A6000",6,48,0.3001,219.07,[143,173],768,77,0.6336,"203.7 tok\u002Fs",203.7,{"effective_value":196,"effective_unit":150,"capacity_factor":155},142.6,[],false,24,"value",[202,206,210,218,225],{"name":184,"vram_gb":186,"cheapest_hour":187,"cost":203,"provider":143,"estimated_tps":194},{"hourly":187,"daily":204,"monthly":188,"yearly":205},7.2,2628.88,{"name":139,"vram_gb":84,"cheapest_hour":140,"cost":207,"provider":143,"estimated_tps":152},{"hourly":140,"daily":208,"monthly":141,"yearly":209},7.73,2822.47,{"name":211,"vram_gb":186,"cheapest_hour":212,"cost":213,"provider":143,"estimated_tps":217},"L40S",0.4669,{"hourly":212,"daily":214,"monthly":215,"yearly":216},11.21,340.84,4090.04,229.2,{"name":219,"vram_gb":186,"cheapest_hour":220,"cost":221,"provider":143,"estimated_tps":38},"RTX PRO 6000 WS",0.8022,{"hourly":220,"daily":222,"monthly":223,"yearly":224},19.25,585.61,7027.27,{"name":226,"vram_gb":186,"cheapest_hour":227,"cost":228,"provider":143,"estimated_tps":38},"RTX PRO 6000 S",1.0131,{"hourly":227,"daily":229,"monthly":230,"yearly":231},24.31,739.56,8874.76,{"posts_count":38,"benchmarks_count":38}]