[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-mistral-medium-35-128b":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":142,"fitting_gpus":214,"community":230},253,"mistral-medium-35-128b","Mistral Medium 3.5 128B","mistralai\u002FMistral-Medium-3.5-128B","mistralai","transformer",null,128000000000,"128B",262144,"MIT",80744,418,"published","https:\u002F\u002Fhuggingface.co\u002Fmistralai\u002FMistral-Medium-3.5-128B","2026-08-12T03:11:19+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},"OCR \u002F Vision","ocr-vision",{"name":27,"slug":28},"Mistral","mistral",0.2547,[],[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":96,"AWQ":100,"GPTQ":114,"GGUF":128},{"minimum":60,"production":69,"recommended":76},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":63,"kv_cache_gb":49,"vram_gb":64,"ram_gb":65,"disk_gb":66,"requirement_source":67,"confidence":68},"minimum",2048,238.42,305.17,457.76,410.93,"estimated",0.95,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":63,"kv_cache_gb":44,"vram_gb":73,"ram_gb":74,"disk_gb":75,"requirement_source":67,"confidence":68},"production",8192,2,350.62,525.93,475.93,{"scenario":77,"context_length":78,"batch_size":40,"weight_gb":63,"kv_cache_gb":40,"vram_gb":79,"ram_gb":80,"disk_gb":81,"requirement_source":67,"confidence":68},"recommended",4096,321.06,481.59,436.93,{"minimum":83,"production":88,"recommended":92},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":84,"kv_cache_gb":49,"vram_gb":85,"ram_gb":86,"disk_gb":87,"requirement_source":67,"confidence":68},119.21,154.37,231.56,224.97,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":84,"kv_cache_gb":44,"vram_gb":89,"ram_gb":90,"disk_gb":91,"requirement_source":67,"confidence":68},186.11,279.16,289.97,{"scenario":77,"context_length":78,"batch_size":40,"weight_gb":84,"kv_cache_gb":40,"vram_gb":93,"ram_gb":94,"disk_gb":95,"requirement_source":67,"confidence":68},163.4,245.11,250.97,{"minimum":97,"production":98,"recommended":99},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":84,"kv_cache_gb":49,"vram_gb":85,"ram_gb":86,"disk_gb":87,"requirement_source":67,"confidence":68},{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":84,"kv_cache_gb":44,"vram_gb":89,"ram_gb":90,"disk_gb":91,"requirement_source":67,"confidence":68},{"scenario":77,"context_length":78,"batch_size":40,"weight_gb":84,"kv_cache_gb":40,"vram_gb":93,"ram_gb":94,"disk_gb":95,"requirement_source":67,"confidence":68},{"minimum":101,"production":106,"recommended":110},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":102,"kv_cache_gb":49,"vram_gb":103,"ram_gb":104,"disk_gb":105,"requirement_source":67,"confidence":68},61.84,81.8,122.7,135.47,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":102,"kv_cache_gb":44,"vram_gb":107,"ram_gb":108,"disk_gb":109,"requirement_source":67,"confidence":68},106.94,160.41,200.47,{"scenario":77,"context_length":78,"batch_size":40,"weight_gb":102,"kv_cache_gb":40,"vram_gb":111,"ram_gb":112,"disk_gb":113,"requirement_source":67,"confidence":68},87.53,131.3,161.47,{"minimum":115,"production":120,"recommended":124},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":116,"kv_cache_gb":49,"vram_gb":117,"ram_gb":118,"disk_gb":119,"requirement_source":67,"confidence":68},62.58,82.74,124.12,136.63,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":116,"kv_cache_gb":44,"vram_gb":121,"ram_gb":122,"disk_gb":123,"requirement_source":67,"confidence":68},107.97,161.95,201.63,{"scenario":77,"context_length":78,"batch_size":40,"weight_gb":116,"kv_cache_gb":40,"vram_gb":125,"ram_gb":126,"disk_gb":127,"requirement_source":67,"confidence":68},88.52,132.78,162.63,{"minimum":129,"production":134,"recommended":138},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":130,"kv_cache_gb":49,"vram_gb":131,"ram_gb":132,"disk_gb":133,"requirement_source":67,"confidence":68},64.07,84.63,126.94,138.96,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":130,"kv_cache_gb":44,"vram_gb":135,"ram_gb":136,"disk_gb":137,"requirement_source":67,"confidence":68},110.02,165.04,203.96,{"scenario":77,"context_length":78,"batch_size":40,"weight_gb":130,"kv_cache_gb":40,"vram_gb":139,"ram_gb":140,"disk_gb":141,"requirement_source":67,"confidence":68},90.49,135.73,164.96,{"recommended":143,"cheapest":162,"performance":175,"min_complexity":184,"cluster_options":199,"no_match":211,"required_vram_gb":111,"candidate_count":212,"usage":213},{"gpu_model":144,"gpu_model_id":145,"gpu_count":40,"total_vram_gb":146,"hour_price":147,"monthly_cost":148,"providers":149,"bandwidth_gbps":151,"fp16_tflops":152,"interconnect_score":153,"parallel_efficiency":40,"score":154,"tier":155,"perf_metric":156,"perf_label":157,"perf_value":158,"workload":10,"cost_metric":159},"H200 SXM",11,141,3.5,2555,[150],"lambda",4800,989,40,0.6366,"balanced","tok\u002Fs","77.6 tok\u002Fs",77.6,{"effective_value":160,"effective_unit":156,"capacity_factor":161},54.3,0.7,{"gpu_model":163,"gpu_model_id":164,"gpu_count":44,"total_vram_gb":165,"hour_price":166,"monthly_cost":167,"providers":168,"bandwidth_gbps":41,"fp16_tflops":41,"interconnect_score":153,"parallel_efficiency":170,"score":171,"tier":172,"perf_metric":156,"perf_label":173,"perf_value":41,"workload":10,"cost_metric":174},"Tesla V100",17,128,0.0272,158.85,[169],"vast",0.4,0.539,"cheapest","0 tok\u002Fs",{"effective_value":41,"effective_unit":156,"capacity_factor":161},{"gpu_model":144,"gpu_model_id":145,"gpu_count":44,"total_vram_gb":176,"hour_price":147,"monthly_cost":177,"providers":178,"bandwidth_gbps":151,"fp16_tflops":152,"interconnect_score":153,"parallel_efficiency":170,"score":179,"tier":155,"perf_metric":156,"perf_label":180,"perf_value":181,"workload":10,"cost_metric":182},1128,20440,[150],0.2197,"248.4 tok\u002Fs",248.4,{"effective_value":183,"effective_unit":156,"capacity_factor":161},173.9,{"gpu_model":185,"gpu_model_id":186,"gpu_count":72,"total_vram_gb":187,"hour_price":188,"monthly_cost":189,"providers":190,"bandwidth_gbps":191,"fp16_tflops":192,"interconnect_score":153,"parallel_efficiency":193,"score":194,"tier":155,"perf_metric":156,"perf_label":195,"perf_value":196,"workload":10,"cost_metric":197},"RTX A6000",6,96,0.3001,438.15,[169,150],768,77,0.8,0.604,"19.9 tok\u002Fs",19.9,{"effective_value":198,"effective_unit":156,"capacity_factor":161},13.9,[200],{"gpu_model":201,"gpu_model_id":202,"gpu_count":44,"total_vram_gb":203,"hour_price":204,"monthly_cost":205,"providers":206,"bandwidth_gbps":207,"fp16_tflops":202,"interconnect_score":153,"parallel_efficiency":170,"insufficient":208,"vram_shortfall_gb":209,"score":210,"tier":155},"RTX 5060 Ti",24,64,0.0554,323.54,[169],448,true,23.53,0.5102,false,18,"value",[215,219,227],{"name":144,"vram_gb":146,"cheapest_hour":147,"cost":216,"provider":150,"estimated_tps":158},{"hourly":147,"daily":217,"monthly":148,"yearly":218},84,30660,{"name":220,"vram_gb":221,"cheapest_hour":222,"cost":223,"provider":169,"estimated_tps":41},"B200",180,4.1271,{"hourly":222,"daily":224,"monthly":225,"yearly":226},99.05,3012.78,36153.4,{"name":228,"vram_gb":221,"cheapest_hour":10,"cost":10,"provider":10,"estimated_tps":229},"B200 SXM",126.1,{"posts_count":41,"benchmarks_count":41}]