[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"model-yi-vl-34b":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":141,"fitting_gpus":209,"community":244},407,"yi-vl-34b","Yi-VL 34B","01-ai\u002FYi-VL-34B","01-ai","transformer",null,34000000000,"34B",4096,"MIT",221,265,"published","https:\u002F\u002Fhuggingface.co\u002F01-ai\u002FYi-VL-34B","2026-08-12T03:11:27+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},"Yi","yi",0.0029,[],[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":81,"INT8":95,"AWQ":99,"GPTQ":113,"GGUF":127},{"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",2048,63.33,83.69,125.53,137.79,"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",8192,2,109,163.49,202.79,{"scenario":77,"context_length":13,"batch_size":40,"weight_gb":63,"kv_cache_gb":40,"vram_gb":78,"ram_gb":79,"disk_gb":80,"requirement_source":67,"confidence":68},"recommended",89.5,134.26,163.79,{"minimum":82,"production":87,"recommended":91},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":83,"kv_cache_gb":49,"vram_gb":84,"ram_gb":85,"disk_gb":86,"requirement_source":67,"confidence":68},31.66,43.63,65.45,88.4,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":83,"kv_cache_gb":44,"vram_gb":88,"ram_gb":89,"disk_gb":90,"requirement_source":67,"confidence":68},65.3,97.95,153.4,{"scenario":77,"context_length":13,"batch_size":40,"weight_gb":83,"kv_cache_gb":40,"vram_gb":92,"ram_gb":93,"disk_gb":94,"requirement_source":67,"confidence":68},47.63,71.44,114.4,{"minimum":96,"production":97,"recommended":98},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":83,"kv_cache_gb":49,"vram_gb":84,"ram_gb":85,"disk_gb":86,"requirement_source":67,"confidence":68},{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":83,"kv_cache_gb":44,"vram_gb":88,"ram_gb":89,"disk_gb":90,"requirement_source":67,"confidence":68},{"scenario":77,"context_length":13,"batch_size":40,"weight_gb":83,"kv_cache_gb":40,"vram_gb":92,"ram_gb":93,"disk_gb":94,"requirement_source":67,"confidence":68},{"minimum":100,"production":105,"recommended":109},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":101,"kv_cache_gb":49,"vram_gb":102,"ram_gb":103,"disk_gb":104,"requirement_source":67,"confidence":68},16.43,24.35,36.53,64.62,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":101,"kv_cache_gb":44,"vram_gb":106,"ram_gb":107,"disk_gb":108,"requirement_source":67,"confidence":68},44.27,66.4,129.62,{"scenario":77,"context_length":13,"batch_size":40,"weight_gb":101,"kv_cache_gb":40,"vram_gb":110,"ram_gb":111,"disk_gb":112,"requirement_source":67,"confidence":68},27.47,41.21,90.62,{"minimum":114,"production":119,"recommended":123},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":115,"kv_cache_gb":49,"vram_gb":116,"ram_gb":117,"disk_gb":118,"requirement_source":67,"confidence":68},16.62,24.6,36.91,64.93,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":115,"kv_cache_gb":44,"vram_gb":120,"ram_gb":121,"disk_gb":122,"requirement_source":67,"confidence":68},44.54,66.81,129.93,{"scenario":77,"context_length":13,"batch_size":40,"weight_gb":115,"kv_cache_gb":40,"vram_gb":124,"ram_gb":125,"disk_gb":126,"requirement_source":67,"confidence":68},27.74,41.6,90.93,{"minimum":128,"production":133,"recommended":137},{"scenario":61,"context_length":62,"batch_size":40,"weight_gb":129,"kv_cache_gb":49,"vram_gb":130,"ram_gb":131,"disk_gb":132,"requirement_source":67,"confidence":68},17.02,25.11,37.66,65.55,{"scenario":70,"context_length":71,"batch_size":72,"weight_gb":129,"kv_cache_gb":44,"vram_gb":134,"ram_gb":135,"disk_gb":136,"requirement_source":67,"confidence":68},45.09,67.63,130.55,{"scenario":77,"context_length":13,"batch_size":40,"weight_gb":129,"kv_cache_gb":40,"vram_gb":138,"ram_gb":139,"disk_gb":140,"requirement_source":67,"confidence":68},28.26,42.39,91.55,{"recommended":142,"cheapest":164,"performance":176,"min_complexity":191,"cluster_options":205,"no_match":206,"required_vram_gb":110,"candidate_count":207,"usage":208},{"gpu_model":143,"gpu_model_id":144,"gpu_count":40,"total_vram_gb":145,"hour_price":146,"monthly_cost":147,"providers":148,"bandwidth_gbps":153,"fp16_tflops":154,"interconnect_score":155,"parallel_efficiency":40,"score":156,"tier":157,"perf_metric":158,"perf_label":159,"perf_value":160,"workload":10,"cost_metric":161},"H100 SXM",10,80,1.6022,1169.61,[149,150,151,152],"vast","tensordock","lambda","runpod",3350,66.9,40,0.5811,"balanced","tok\u002Fs","203.9 tok\u002Fs",203.9,{"effective_value":162,"effective_unit":158,"capacity_factor":163},142.7,0.7,{"gpu_model":165,"gpu_model_id":166,"gpu_count":72,"total_vram_gb":167,"hour_price":168,"monthly_cost":169,"providers":170,"bandwidth_gbps":41,"fp16_tflops":41,"interconnect_score":155,"parallel_efficiency":171,"score":172,"tier":173,"perf_metric":158,"perf_label":174,"perf_value":41,"workload":10,"cost_metric":175},"Tesla V100",17,32,0.0272,39.71,[149],0.8,0.6178,"cheapest","0 tok\u002Fs",{"effective_value":41,"effective_unit":158,"capacity_factor":163},{"gpu_model":177,"gpu_model_id":178,"gpu_count":44,"total_vram_gb":179,"hour_price":180,"monthly_cost":181,"providers":182,"bandwidth_gbps":183,"fp16_tflops":184,"interconnect_score":155,"parallel_efficiency":185,"score":186,"tier":157,"perf_metric":158,"perf_label":187,"perf_value":188,"workload":10,"cost_metric":189},"H200 SXM",11,1128,3.5,20440,[151],4800,989,0.4,0.2197,"934.9 tok\u002Fs",934.9,{"effective_value":190,"effective_unit":158,"capacity_factor":163},654.4,{"gpu_model":192,"gpu_model_id":193,"gpu_count":40,"total_vram_gb":194,"hour_price":195,"monthly_cost":196,"providers":197,"bandwidth_gbps":198,"fp16_tflops":199,"interconnect_score":155,"parallel_efficiency":40,"score":200,"tier":173,"perf_metric":158,"perf_label":201,"perf_value":202,"workload":10,"cost_metric":203},"RTX A6000",6,48,0.3001,219.07,[149,151],768,77,0.6435,"46.7 tok\u002Fs",46.7,{"effective_value":204,"effective_unit":158,"capacity_factor":163},32.7,[],false,24,"value",[210,214,222,230,237],{"name":192,"vram_gb":194,"cheapest_hour":195,"cost":211,"provider":149,"estimated_tps":202},{"hourly":195,"daily":212,"monthly":196,"yearly":213},7.2,2628.88,{"name":215,"vram_gb":167,"cheapest_hour":216,"cost":217,"provider":149,"estimated_tps":221},"RTX 5090",0.3222,{"hourly":216,"daily":218,"monthly":219,"yearly":220},7.73,235.21,2822.47,109.1,{"name":223,"vram_gb":194,"cheapest_hour":224,"cost":225,"provider":149,"estimated_tps":229},"L40S",0.4669,{"hourly":224,"daily":226,"monthly":227,"yearly":228},11.21,340.84,4090.04,52.6,{"name":231,"vram_gb":194,"cheapest_hour":232,"cost":233,"provider":149,"estimated_tps":41},"RTX PRO 6000 WS",0.8022,{"hourly":232,"daily":234,"monthly":235,"yearly":236},19.25,585.61,7027.27,{"name":238,"vram_gb":194,"cheapest_hour":239,"cost":240,"provider":149,"estimated_tps":41},"RTX PRO 6000 S",1.0131,{"hourly":239,"daily":241,"monthly":242,"yearly":243},24.31,739.56,8874.76,{"posts_count":41,"benchmarks_count":41}]