[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-glm-45-air":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":18,"huggingface_url":19,"created_at":20,"seo":21,"category":23,"family":26,"score":29,"variants":30,"quantizations":31,"requirements":58,"gpu_recommendations":143,"fitting_gpus":213,"community":229},12,"glm-45-air","GLM-4.5 Air","zai-org\u002FGLM-4.5-Air","zai-org","moe_transformer",null,106000000000,"106B",131072,"MIT",210000,1100,"published","GLM-4.5 Air，边缘优化的高效版本。","https:\u002F\u002Fhuggingface.co\u002Fzai-org\u002FGLM-4.5-Air","2026-08-12T02:22:23+00:00",{"title":6,"description":18,"image":10,"robots":22,"canonical_url":10},"index, follow",{"name":24,"slug":25},"LLM","llm",{"name":27,"slug":28},"GLM","glm",0.2625,[],[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":83,"INT8":97,"AWQ":101,"GPTQ":115,"GGUF":129},{"minimum":60,"production":69,"recommended":76},{"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",4096,197.44,254.71,382.07,347.01,"estimated",0.95,{"scenario":70,"context_length":71,"batch_size":34,"weight_gb":63,"kv_cache_gb":72,"vram_gb":73,"ram_gb":74,"disk_gb":75,"requirement_source":67,"confidence":68},"production",16384,64,390.07,585.1,412.01,{"scenario":77,"context_length":78,"batch_size":79,"weight_gb":63,"kv_cache_gb":44,"vram_gb":80,"ram_gb":81,"disk_gb":82,"requirement_source":67,"confidence":68},"recommended",8192,2,281.81,422.72,373.01,{"minimum":84,"production":89,"recommended":93},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":85,"kv_cache_gb":45,"vram_gb":86,"ram_gb":87,"disk_gb":88,"requirement_source":67,"confidence":68},98.72,129.83,194.75,193,{"scenario":70,"context_length":71,"batch_size":34,"weight_gb":85,"kv_cache_gb":72,"vram_gb":90,"ram_gb":91,"disk_gb":92,"requirement_source":67,"confidence":68},253.83,380.75,258,{"scenario":77,"context_length":78,"batch_size":79,"weight_gb":85,"kv_cache_gb":44,"vram_gb":94,"ram_gb":95,"disk_gb":96,"requirement_source":67,"confidence":68},151.26,226.89,219,{"minimum":98,"production":99,"recommended":100},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":85,"kv_cache_gb":45,"vram_gb":86,"ram_gb":87,"disk_gb":88,"requirement_source":67,"confidence":68},{"scenario":70,"context_length":71,"batch_size":34,"weight_gb":85,"kv_cache_gb":72,"vram_gb":90,"ram_gb":91,"disk_gb":92,"requirement_source":67,"confidence":68},{"scenario":77,"context_length":78,"batch_size":79,"weight_gb":85,"kv_cache_gb":44,"vram_gb":94,"ram_gb":95,"disk_gb":96,"requirement_source":67,"confidence":68},{"minimum":102,"production":107,"recommended":111},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":103,"kv_cache_gb":45,"vram_gb":104,"ram_gb":105,"disk_gb":106,"requirement_source":67,"confidence":68},51.21,69.73,104.6,118.89,{"scenario":70,"context_length":71,"batch_size":34,"weight_gb":103,"kv_cache_gb":72,"vram_gb":108,"ram_gb":109,"disk_gb":110,"requirement_source":67,"confidence":68},188.27,282.41,183.89,{"scenario":77,"context_length":78,"batch_size":79,"weight_gb":103,"kv_cache_gb":44,"vram_gb":112,"ram_gb":113,"disk_gb":114,"requirement_source":67,"confidence":68},88.43,132.64,144.89,{"minimum":116,"production":121,"recommended":125},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":117,"kv_cache_gb":45,"vram_gb":118,"ram_gb":119,"disk_gb":120,"requirement_source":67,"confidence":68},51.83,70.51,105.77,119.85,{"scenario":70,"context_length":71,"batch_size":34,"weight_gb":117,"kv_cache_gb":72,"vram_gb":122,"ram_gb":123,"disk_gb":124,"requirement_source":67,"confidence":68},189.12,283.68,184.85,{"scenario":77,"context_length":78,"batch_size":79,"weight_gb":117,"kv_cache_gb":44,"vram_gb":126,"ram_gb":127,"disk_gb":128,"requirement_source":67,"confidence":68},89.24,133.86,145.85,{"minimum":130,"production":135,"recommended":139},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":131,"kv_cache_gb":45,"vram_gb":132,"ram_gb":133,"disk_gb":134,"requirement_source":67,"confidence":68},53.06,72.07,108.11,121.78,{"scenario":70,"context_length":71,"batch_size":34,"weight_gb":131,"kv_cache_gb":72,"vram_gb":136,"ram_gb":137,"disk_gb":138,"requirement_source":67,"confidence":68},190.83,286.24,186.78,{"scenario":77,"context_length":78,"batch_size":79,"weight_gb":131,"kv_cache_gb":44,"vram_gb":140,"ram_gb":141,"disk_gb":142,"requirement_source":67,"confidence":68},90.87,136.31,147.78,{"recommended":144,"cheapest":164,"performance":181,"min_complexity":190,"cluster_options":199,"no_match":210,"required_vram_gb":112,"candidate_count":211,"usage":212},{"gpu_model":145,"gpu_model_id":146,"gpu_count":34,"total_vram_gb":147,"hour_price":148,"monthly_cost":149,"providers":150,"bandwidth_gbps":152,"fp16_tflops":153,"interconnect_score":154,"parallel_efficiency":155,"score":156,"tier":157,"perf_metric":158,"perf_label":159,"perf_value":160,"workload":10,"cost_metric":161},"H200 SXM",11,564,3.5,10220,[151],"lambda",4800,989,40,0.6,0.3112,"balanced","tok\u002Fs","225 tok\u002Fs",225,{"effective_value":162,"effective_unit":158,"capacity_factor":163},157.5,0.7,{"gpu_model":165,"gpu_model_id":166,"gpu_count":44,"total_vram_gb":167,"hour_price":168,"monthly_cost":169,"providers":170,"bandwidth_gbps":172,"fp16_tflops":173,"interconnect_score":154,"parallel_efficiency":174,"score":175,"tier":176,"perf_metric":158,"perf_label":177,"perf_value":178,"workload":10,"cost_metric":179},"RTX 5070",28,96,0.0101,58.98,[171],"vast",672,61,0.4,0.5285,"cheapest","42 tok\u002Fs",42,{"effective_value":180,"effective_unit":158,"capacity_factor":163},29.4,{"gpu_model":145,"gpu_model_id":146,"gpu_count":44,"total_vram_gb":182,"hour_price":148,"monthly_cost":183,"providers":184,"bandwidth_gbps":152,"fp16_tflops":153,"interconnect_score":154,"parallel_efficiency":174,"score":185,"tier":157,"perf_metric":158,"perf_label":186,"perf_value":187,"workload":10,"cost_metric":188},1128,20440,[151],0.2197,"299.9 tok\u002Fs",299.9,{"effective_value":189,"effective_unit":158,"capacity_factor":163},209.9,{"gpu_model":145,"gpu_model_id":146,"gpu_count":40,"total_vram_gb":191,"hour_price":148,"monthly_cost":192,"providers":193,"bandwidth_gbps":152,"fp16_tflops":153,"interconnect_score":154,"parallel_efficiency":40,"score":194,"tier":157,"perf_metric":158,"perf_label":195,"perf_value":196,"workload":10,"cost_metric":197},141,2555,[151],0.6383,"93.7 tok\u002Fs",93.7,{"effective_value":198,"effective_unit":158,"capacity_factor":163},65.6,[200],{"gpu_model":201,"gpu_model_id":202,"gpu_count":44,"total_vram_gb":72,"hour_price":203,"monthly_cost":204,"providers":205,"bandwidth_gbps":206,"fp16_tflops":202,"interconnect_score":154,"parallel_efficiency":174,"insufficient":207,"vram_shortfall_gb":208,"score":209,"tier":157},"RTX 5060 Ti",24,0.0544,317.7,[171],448,true,24.43,0.5103,false,17,"value",[214,218,226],{"name":145,"vram_gb":191,"cheapest_hour":148,"cost":215,"provider":151,"estimated_tps":196},{"hourly":148,"daily":216,"monthly":192,"yearly":217},84,30660,{"name":219,"vram_gb":220,"cheapest_hour":221,"cost":222,"provider":171,"estimated_tps":41},"B200",180,4.1271,{"hourly":221,"daily":223,"monthly":224,"yearly":225},99.05,3012.78,36153.4,{"name":227,"vram_gb":220,"cheapest_hour":10,"cost":10,"provider":10,"estimated_tps":228},"B200 SXM",152.3,{"posts_count":41,"benchmarks_count":41}]