[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f2zjqgk5o6jc26":3,"$f2feu3wuop23tq":92,"$f224kinx8zv2ho":94,"$f3bmd3rntvzzx1":532,"$f8ypa021er9qz":546,"$f3p3vq051jld6o":548},{"id":4,"name":5,"publisher":6,"paramsB":7,"useCase":8,"licenseStatus":9,"licenseUrl":10,"sourceUrl":10,"descriptionZh":10,"descriptionEn":10,"imageUrl":10,"downloadUrls":11,"stats":12,"variants":19,"derivations":20,"measuredHardware":23,"updatedAt":91},71,"DeepSeek-R1-Distill-Qwen-14B","DeepSeek",14,"REASONING","UNVERIFIED",null,[],{"recordCount":13,"variantCount":14,"hardwareCount":15,"unlinkedRecordCount":13,"evidence":16},83,0,25,{"L0":17,"L1":14,"L2":18},16,67,[],{"parents":21,"children":22},[],[],[24,28,32,35,39,42,44,47,50,53,55,58,60,63,66,68,71,73,75,77,80,83,85,87,89],{"name":25,"vramGb":26,"recordCount":27},"NVIDIA RTX 3090",24,9,{"name":29,"vramGb":30,"recordCount":31},"NVIDIA RTX 5090",32,8,{"name":33,"vramGb":26,"recordCount":34},"NVIDIA RTX 4090",7,{"name":36,"vramGb":37,"recordCount":38},"Apple M5 Max",128,5,{"name":40,"vramGb":26,"recordCount":41},"AMD Radeon RX 7900 XTX",4,{"name":43,"vramGb":37,"recordCount":41},"AMD Ryzen AI Max+ 395",{"name":45,"vramGb":46,"recordCount":41},"NVIDIA H100 80GB HBM3",80,{"name":48,"vramGb":49,"recordCount":41},"NVIDIA H20",97,{"name":51,"vramGb":52,"recordCount":41},"NVIDIA RTX 5070",12,{"name":54,"vramGb":37,"recordCount":41},"NVIDIA RTX Spark",{"name":56,"vramGb":30,"recordCount":57},"AMD Radeon AI PRO R9700",3,{"name":59,"vramGb":10,"recordCount":57},"Apple M2 Ultra",{"name":61,"vramGb":62,"recordCount":57},"Apple M4 Pro",64,{"name":64,"vramGb":65,"recordCount":57},"Apple M5 Ultra",512,{"name":67,"vramGb":37,"recordCount":57},"NVIDIA DGX Spark",{"name":69,"vramGb":17,"recordCount":70},"AMD Radeon RX 9070 XT",2,{"name":72,"vramGb":62,"recordCount":70},"NVIDIA Jetson AGX Orin 64GB",{"name":74,"vramGb":17,"recordCount":70},"NVIDIA RTX 4080 Super",{"name":76,"vramGb":17,"recordCount":70},"NVIDIA RTX 5080",{"name":78,"vramGb":79,"recordCount":70},"NVIDIA RTX PRO 6000 Blackwell 96GB",96,{"name":81,"vramGb":17,"recordCount":82},"AMD Radeon RX 7800 XT",1,{"name":84,"vramGb":17,"recordCount":82},"AMD Radeon RX 9070",{"name":86,"vramGb":62,"recordCount":82},"Apple M5 Pro",{"name":88,"vramGb":17,"recordCount":82},"NVIDIA RTX 4060 Ti 16GB",{"name":90,"vramGb":17,"recordCount":82},"NVIDIA RTX 4080","2026-10-02T04:13:10.959Z",{"favorited":93},false,{"items":95,"total":531,"page":82,"pageSize":220},[96,112,124,133,144,155,166,177,185,195,205,215,223,232,241,250,261,271,277,285,296,304,314,322,330,337,346,354,363,372,379,386,394,402,408,419,427,434,441,447,454,461,469,476,484,492,499,508,515,523],{"modelId":4,"modelName":5,"hardwareId":82,"hardwareName":33,"frameworkId":82,"frameworkName":97,"quantization":98,"hardwareVramGb":26,"decodeTps":99,"prefillTps":100,"vramGb":101,"genSeconds":10,"modelUseCase":8,"ttftS":102,"ttfbGb":103,"mtpAcceptanceRate":10,"powerW":104,"sampleCount":82,"reproduceCount":82,"sourceTypes":105,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":109,"latestOs":110,"latestCudaVersion":111,"latestFlashAttention":93,"latestOutputSpec":10},"llama.cpp","Q3_K_M",122.06,409.59,22.09,2.196,0.98,306.5,[106],"OTHER","L2","2026-10-19T12:39:43.326Z","v0.47","Ubuntu 24.04","12.8",{"modelId":4,"modelName":5,"hardwareId":70,"hardwareName":25,"frameworkId":82,"frameworkName":97,"quantization":113,"hardwareVramGb":26,"decodeTps":114,"prefillTps":115,"vramGb":116,"genSeconds":10,"modelUseCase":8,"ttftS":117,"ttfbGb":118,"mtpAcceptanceRate":10,"powerW":119,"sampleCount":82,"reproduceCount":82,"sourceTypes":120,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":121,"latestOs":122,"latestCudaVersion":123,"latestFlashAttention":93,"latestOutputSpec":10},"IQ4_XS",79.26,346.57,22.8,2.499,0.63,343.1,[106],"v0.42","Windows 11 24H2","12.6",{"modelId":4,"modelName":5,"hardwareId":70,"hardwareName":25,"frameworkId":82,"frameworkName":97,"quantization":98,"hardwareVramGb":26,"decodeTps":125,"prefillTps":126,"vramGb":101,"genSeconds":10,"modelUseCase":8,"ttftS":127,"ttfbGb":128,"mtpAcceptanceRate":10,"powerW":129,"sampleCount":82,"reproduceCount":82,"sourceTypes":130,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":131,"latestOs":122,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},75.68,415.45,1.008,0.61,293.4,[106],"v0.43",true,{"modelId":4,"modelName":5,"hardwareId":70,"hardwareName":25,"frameworkId":70,"frameworkName":134,"quantization":98,"hardwareVramGb":26,"decodeTps":135,"prefillTps":136,"vramGb":101,"genSeconds":10,"modelUseCase":8,"ttftS":137,"ttfbGb":138,"mtpAcceptanceRate":10,"powerW":139,"sampleCount":82,"reproduceCount":82,"sourceTypes":140,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":141,"latestOs":142,"latestCudaVersion":143,"latestFlashAttention":132,"latestOutputSpec":10},"vLLM",80.11,318.86,0.675,0.64,336.2,[106],"v0.31","Windows 11 23H2","12.4",{"modelId":4,"modelName":5,"hardwareId":145,"hardwareName":40,"frameworkId":82,"frameworkName":97,"quantization":146,"hardwareVramGb":26,"decodeTps":147,"prefillTps":148,"vramGb":116,"genSeconds":10,"modelUseCase":8,"ttftS":149,"ttfbGb":150,"mtpAcceptanceRate":10,"powerW":151,"sampleCount":82,"reproduceCount":82,"sourceTypes":152,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":153,"latestOs":154,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},11,"Q4_K_M",65.83,346.39,2.3,0.53,239.7,[106],"v0.45","Ubuntu 22.04",{"modelId":4,"modelName":5,"hardwareId":145,"hardwareName":40,"frameworkId":57,"frameworkName":156,"quantization":157,"hardwareVramGb":26,"decodeTps":158,"prefillTps":159,"vramGb":160,"genSeconds":10,"modelUseCase":8,"ttftS":161,"ttfbGb":162,"mtpAcceptanceRate":10,"powerW":163,"sampleCount":82,"reproduceCount":82,"sourceTypes":164,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":165,"latestOs":142,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},"Ollama","IQ2_XS",90.76,372.69,16.21,1.026,0.73,249.6,[106],"v0.54",{"modelId":4,"modelName":5,"hardwareId":167,"hardwareName":29,"frameworkId":70,"frameworkName":134,"quantization":168,"hardwareVramGb":30,"decodeTps":169,"prefillTps":170,"vramGb":171,"genSeconds":10,"modelUseCase":8,"ttftS":172,"ttfbGb":173,"mtpAcceptanceRate":10,"powerW":174,"sampleCount":82,"reproduceCount":82,"sourceTypes":175,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":176,"latestOs":110,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},13,"Q8_0",145.18,478.54,30.4,1.694,1.16,423.2,[106],"v0.36",{"modelId":4,"modelName":5,"hardwareId":7,"hardwareName":76,"frameworkId":70,"frameworkName":134,"quantization":157,"hardwareVramGb":17,"decodeTps":178,"prefillTps":179,"vramGb":180,"genSeconds":10,"modelUseCase":8,"ttftS":181,"ttfbGb":182,"mtpAcceptanceRate":10,"powerW":183,"sampleCount":82,"reproduceCount":82,"sourceTypes":184,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":141,"latestOs":142,"latestCudaVersion":143,"latestFlashAttention":93,"latestOutputSpec":10},227.28,1043.49,15.2,0.18,1.82,334.2,[106],{"modelId":4,"modelName":5,"hardwareId":186,"hardwareName":74,"frameworkId":82,"frameworkName":97,"quantization":187,"hardwareVramGb":17,"decodeTps":188,"prefillTps":189,"vramGb":190,"genSeconds":10,"modelUseCase":8,"ttftS":191,"ttfbGb":192,"mtpAcceptanceRate":10,"powerW":193,"sampleCount":82,"reproduceCount":82,"sourceTypes":194,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":141,"latestOs":142,"latestCudaVersion":143,"latestFlashAttention":132,"latestOutputSpec":10},20,"IQ1_S",208.08,731.51,14.45,1.092,1.66,207.1,[106],{"modelId":4,"modelName":5,"hardwareId":26,"hardwareName":43,"frameworkId":70,"frameworkName":134,"quantization":196,"hardwareVramGb":37,"decodeTps":197,"prefillTps":198,"vramGb":199,"genSeconds":10,"modelUseCase":8,"ttftS":200,"ttfbGb":201,"mtpAcceptanceRate":10,"powerW":202,"sampleCount":82,"reproduceCount":82,"sourceTypes":203,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":204,"latestOs":154,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},"Q3_K_S",34.77,161.9,21.21,0.751,0.28,67.1,[106],"v0.44",{"modelId":4,"modelName":5,"hardwareId":26,"hardwareName":43,"frameworkId":70,"frameworkName":134,"quantization":206,"hardwareVramGb":37,"decodeTps":207,"prefillTps":208,"vramGb":209,"genSeconds":10,"modelUseCase":8,"ttftS":210,"ttfbGb":211,"mtpAcceptanceRate":10,"powerW":212,"sampleCount":82,"reproduceCount":82,"sourceTypes":213,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":214,"latestOs":142,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},"Q4_K_S",29.55,96.92,25.03,2.134,0.24,49.7,[106],"v0.26",{"modelId":4,"modelName":5,"hardwareId":26,"hardwareName":43,"frameworkId":57,"frameworkName":156,"quantization":206,"hardwareVramGb":37,"decodeTps":216,"prefillTps":217,"vramGb":209,"genSeconds":10,"modelUseCase":8,"ttftS":218,"ttfbGb":219,"mtpAcceptanceRate":10,"powerW":220,"sampleCount":82,"reproduceCount":82,"sourceTypes":221,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":222,"latestOs":142,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},28.23,99.22,1.422,0.23,50,[106],"v0.49",{"modelId":4,"modelName":5,"hardwareId":224,"hardwareName":36,"frameworkId":82,"frameworkName":97,"quantization":98,"hardwareVramGb":37,"decodeTps":225,"prefillTps":226,"vramGb":101,"genSeconds":10,"modelUseCase":8,"ttftS":227,"ttfbGb":150,"mtpAcceptanceRate":10,"powerW":228,"sampleCount":82,"reproduceCount":82,"sourceTypes":229,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":230,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},28,66.61,292.99,1.371,40.3,[106],"v0.22","macOS 26.1",{"modelId":4,"modelName":5,"hardwareId":224,"hardwareName":36,"frameworkId":57,"frameworkName":156,"quantization":113,"hardwareVramGb":37,"decodeTps":233,"prefillTps":234,"vramGb":235,"genSeconds":10,"modelUseCase":8,"ttftS":236,"ttfbGb":237,"mtpAcceptanceRate":10,"powerW":238,"sampleCount":82,"reproduceCount":82,"sourceTypes":239,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":240,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},63.06,323.43,23.86,0.806,0.5,51.9,[106],"v0.39",{"modelId":4,"modelName":5,"hardwareId":224,"hardwareName":36,"frameworkId":57,"frameworkName":156,"quantization":168,"hardwareVramGb":37,"decodeTps":242,"prefillTps":243,"vramGb":244,"genSeconds":10,"modelUseCase":8,"ttftS":245,"ttfbGb":246,"mtpAcceptanceRate":10,"powerW":247,"sampleCount":82,"reproduceCount":82,"sourceTypes":248,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":249,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":93,"latestOutputSpec":10},36.84,127.42,35.91,0.301,0.29,47.5,[106],"v0.20",{"modelId":4,"modelName":5,"hardwareId":251,"hardwareName":64,"frameworkId":82,"frameworkName":97,"quantization":252,"hardwareVramGb":65,"decodeTps":253,"prefillTps":254,"vramGb":255,"genSeconds":10,"modelUseCase":8,"ttftS":256,"ttfbGb":257,"mtpAcceptanceRate":10,"powerW":258,"sampleCount":82,"reproduceCount":82,"sourceTypes":259,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":260,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},29,"Q5_K_M",77.61,409.9,29.44,1.995,0.62,121.7,[106],"v0.30",{"modelId":4,"modelName":5,"hardwareId":251,"hardwareName":64,"frameworkId":57,"frameworkName":156,"quantization":262,"hardwareVramGb":65,"decodeTps":263,"prefillTps":264,"vramGb":265,"genSeconds":10,"modelUseCase":8,"ttftS":266,"ttfbGb":267,"mtpAcceptanceRate":10,"powerW":268,"sampleCount":82,"reproduceCount":82,"sourceTypes":269,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":270,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},"Q2_K",118.32,453.53,18.27,0.834,0.95,92.4,[106],"v0.24",{"modelId":4,"modelName":5,"hardwareId":251,"hardwareName":64,"frameworkId":57,"frameworkName":156,"quantization":252,"hardwareVramGb":65,"decodeTps":272,"prefillTps":273,"vramGb":255,"genSeconds":10,"modelUseCase":8,"ttftS":274,"ttfbGb":150,"mtpAcceptanceRate":10,"powerW":275,"sampleCount":82,"reproduceCount":82,"sourceTypes":276,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":270,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},65.66,313.69,0.303,116.1,[106],{"modelId":4,"modelName":5,"hardwareId":278,"hardwareName":45,"frameworkId":70,"frameworkName":134,"quantization":113,"hardwareVramGb":46,"decodeTps":279,"prefillTps":280,"vramGb":235,"genSeconds":10,"modelUseCase":8,"ttftS":210,"ttfbGb":281,"mtpAcceptanceRate":10,"powerW":282,"sampleCount":82,"reproduceCount":82,"sourceTypes":283,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":284,"latestOs":154,"latestCudaVersion":111,"latestFlashAttention":93,"latestOutputSpec":10},55,398.24,1414.76,3.19,557.8,[106],"v0.25",{"modelId":4,"modelName":5,"hardwareId":278,"hardwareName":45,"frameworkId":70,"frameworkName":134,"quantization":286,"hardwareVramGb":46,"decodeTps":287,"prefillTps":288,"vramGb":289,"genSeconds":10,"modelUseCase":8,"ttftS":290,"ttfbGb":291,"mtpAcceptanceRate":10,"powerW":292,"sampleCount":82,"reproduceCount":82,"sourceTypes":293,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":294,"latestOs":154,"latestCudaVersion":295,"latestFlashAttention":132,"latestOutputSpec":10},"Q5_K_S",364.79,1787.53,28.56,2.337,2.92,671.3,[106],"v0.53","13.0",{"modelId":4,"modelName":5,"hardwareId":278,"hardwareName":45,"frameworkId":57,"frameworkName":156,"quantization":157,"hardwareVramGb":46,"decodeTps":297,"prefillTps":298,"vramGb":160,"genSeconds":10,"modelUseCase":8,"ttftS":299,"ttfbGb":300,"mtpAcceptanceRate":10,"powerW":301,"sampleCount":82,"reproduceCount":82,"sourceTypes":302,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":303,"latestOs":110,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},291.58,1457.4,2.205,2.33,601.5,[106],"v0.40",{"modelId":4,"modelName":5,"hardwareId":278,"hardwareName":45,"frameworkId":57,"frameworkName":156,"quantization":305,"hardwareVramGb":46,"decodeTps":306,"prefillTps":307,"vramGb":308,"genSeconds":10,"modelUseCase":8,"ttftS":309,"ttfbGb":310,"mtpAcceptanceRate":10,"powerW":311,"sampleCount":82,"reproduceCount":82,"sourceTypes":312,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":313,"latestOs":110,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},"IQ3_XS",340.5,1221.18,20.33,1.729,2.72,647.6,[106],"v0.23",{"modelId":4,"modelName":5,"hardwareId":315,"hardwareName":48,"frameworkId":82,"frameworkName":97,"quantization":157,"hardwareVramGb":49,"decodeTps":316,"prefillTps":317,"vramGb":160,"genSeconds":10,"modelUseCase":8,"ttftS":318,"ttfbGb":319,"mtpAcceptanceRate":10,"powerW":320,"sampleCount":82,"reproduceCount":82,"sourceTypes":321,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":153,"latestOs":122,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},56,67.98,225.25,1.774,0.54,339.3,[106],{"modelId":4,"modelName":5,"hardwareId":315,"hardwareName":48,"frameworkId":82,"frameworkName":97,"quantization":146,"hardwareVramGb":49,"decodeTps":323,"prefillTps":324,"vramGb":325,"genSeconds":10,"modelUseCase":8,"ttftS":326,"ttfbGb":327,"mtpAcceptanceRate":10,"powerW":328,"sampleCount":82,"reproduceCount":82,"sourceTypes":329,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":153,"latestOs":110,"latestCudaVersion":111,"latestFlashAttention":93,"latestOutputSpec":10},42.42,221.34,26.21,0.144,0.34,303.6,[106],{"modelId":4,"modelName":5,"hardwareId":315,"hardwareName":48,"frameworkId":70,"frameworkName":134,"quantization":262,"hardwareVramGb":49,"decodeTps":331,"prefillTps":332,"vramGb":265,"genSeconds":10,"modelUseCase":8,"ttftS":333,"ttfbGb":237,"mtpAcceptanceRate":10,"powerW":334,"sampleCount":82,"reproduceCount":82,"sourceTypes":335,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":336,"latestOs":122,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},62.74,220.68,0.256,350.9,[106],"v0.21",{"modelId":4,"modelName":5,"hardwareId":315,"hardwareName":48,"frameworkId":70,"frameworkName":134,"quantization":338,"hardwareVramGb":49,"decodeTps":339,"prefillTps":340,"vramGb":341,"genSeconds":10,"modelUseCase":8,"ttftS":342,"ttfbGb":343,"mtpAcceptanceRate":10,"powerW":344,"sampleCount":82,"reproduceCount":82,"sourceTypes":345,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":230,"latestOs":110,"latestCudaVersion":143,"latestFlashAttention":93,"latestOutputSpec":10},"Q6_K",33.53,137.11,31.5,1.578,0.27,382.5,[106],{"modelId":4,"modelName":5,"hardwareId":347,"hardwareName":78,"frameworkId":82,"frameworkName":97,"quantization":286,"hardwareVramGb":79,"decodeTps":348,"prefillTps":349,"vramGb":289,"genSeconds":10,"modelUseCase":8,"ttftS":350,"ttfbGb":351,"mtpAcceptanceRate":10,"powerW":352,"sampleCount":82,"reproduceCount":82,"sourceTypes":353,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":131,"latestOs":110,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},68,172.78,845.2,2.469,1.38,338.6,[106],{"modelId":4,"modelName":5,"hardwareId":347,"hardwareName":78,"frameworkId":70,"frameworkName":134,"quantization":355,"hardwareVramGb":79,"decodeTps":356,"prefillTps":357,"vramGb":358,"genSeconds":10,"modelUseCase":8,"ttftS":359,"ttfbGb":360,"mtpAcceptanceRate":10,"powerW":361,"sampleCount":82,"reproduceCount":82,"sourceTypes":362,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":294,"latestOs":122,"latestCudaVersion":143,"latestFlashAttention":132,"latestOutputSpec":10},"IQ3_S",236.15,1107.64,21.5,1.276,1.89,340,[106],{"modelId":4,"modelName":5,"hardwareId":364,"hardwareName":59,"frameworkId":82,"frameworkName":97,"quantization":206,"hardwareVramGb":10,"decodeTps":365,"prefillTps":366,"vramGb":209,"genSeconds":10,"modelUseCase":8,"ttftS":367,"ttfbGb":368,"mtpAcceptanceRate":10,"powerW":369,"sampleCount":82,"reproduceCount":82,"sourceTypes":370,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":371,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},103,92.27,397.78,0.421,0.74,130.9,[106],"v0.37",{"modelId":4,"modelName":5,"hardwareId":364,"hardwareName":59,"frameworkId":57,"frameworkName":156,"quantization":196,"hardwareVramGb":10,"decodeTps":373,"prefillTps":374,"vramGb":199,"genSeconds":10,"modelUseCase":8,"ttftS":375,"ttfbGb":376,"mtpAcceptanceRate":10,"powerW":377,"sampleCount":82,"reproduceCount":82,"sourceTypes":378,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":204,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},106.35,407.93,0.881,0.85,149.9,[106],{"modelId":4,"modelName":5,"hardwareId":364,"hardwareName":59,"frameworkId":57,"frameworkName":156,"quantization":252,"hardwareVramGb":10,"decodeTps":380,"prefillTps":381,"vramGb":255,"genSeconds":10,"modelUseCase":8,"ttftS":382,"ttfbGb":150,"mtpAcceptanceRate":10,"powerW":383,"sampleCount":82,"reproduceCount":82,"sourceTypes":384,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":385,"latestOs":231,"latestCudaVersion":10,"latestFlashAttention":132,"latestOutputSpec":10},66.35,310.63,0.559,142.2,[106],"v0.52",{"modelId":4,"modelName":5,"hardwareId":387,"hardwareName":54,"frameworkId":82,"frameworkName":97,"quantization":146,"hardwareVramGb":37,"decodeTps":388,"prefillTps":389,"vramGb":325,"genSeconds":10,"modelUseCase":8,"ttftS":390,"ttfbGb":391,"mtpAcceptanceRate":10,"powerW":380,"sampleCount":70,"reproduceCount":82,"sourceTypes":392,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":393,"latestOs":122,"latestCudaVersion":295,"latestFlashAttention":93,"latestOutputSpec":10},110,32.88,122.17,1.587,0.26,[106],"v0.28",{"modelId":4,"modelName":5,"hardwareId":387,"hardwareName":54,"frameworkId":57,"frameworkName":156,"quantization":355,"hardwareVramGb":37,"decodeTps":395,"prefillTps":396,"vramGb":358,"genSeconds":10,"modelUseCase":8,"ttftS":397,"ttfbGb":398,"mtpAcceptanceRate":10,"powerW":399,"sampleCount":82,"reproduceCount":82,"sourceTypes":400,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":401,"latestOs":110,"latestCudaVersion":295,"latestFlashAttention":93,"latestOutputSpec":10},36.95,186.83,0.342,0.3,54.2,[106],"v0.27",{"modelId":4,"modelName":5,"hardwareId":387,"hardwareName":54,"frameworkId":57,"frameworkName":156,"quantization":146,"hardwareVramGb":37,"decodeTps":403,"prefillTps":404,"vramGb":325,"genSeconds":10,"modelUseCase":8,"ttftS":405,"ttfbGb":391,"mtpAcceptanceRate":10,"powerW":406,"sampleCount":82,"reproduceCount":82,"sourceTypes":407,"evidenceLevel":107,"updatedAt":108,"latestFrameworkVersion":303,"latestOs":110,"latestCudaVersion":295,"latestFlashAttention":93,"latestOutputSpec":10},32.3,137.42,2.064,51.4,[106],{"modelId":4,"modelName":5,"hardwareId":82,"hardwareName":33,"frameworkId":70,"frameworkName":134,"quantization":409,"hardwareVramGb":26,"decodeTps":410,"prefillTps":411,"vramGb":412,"genSeconds":10,"modelUseCase":8,"ttftS":413,"ttfbGb":414,"mtpAcceptanceRate":10,"powerW":415,"sampleCount":82,"reproduceCount":82,"sourceTypes":416,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":418,"latestOs":122,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},"IQ2_M",163.66,743.62,17.39,2.421,1.31,390.2,[106],"2026-10-18T02:53:37.888Z","v0.51",{"modelId":4,"modelName":5,"hardwareId":82,"hardwareName":33,"frameworkId":57,"frameworkName":156,"quantization":420,"hardwareVramGb":26,"decodeTps":421,"prefillTps":422,"vramGb":116,"genSeconds":10,"modelUseCase":8,"ttftS":423,"ttfbGb":424,"mtpAcceptanceRate":10,"powerW":425,"sampleCount":82,"reproduceCount":82,"sourceTypes":426,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":230,"latestOs":122,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},"Q3_K_L",117.29,610.83,2.207,0.94,339.9,[106],{"modelId":4,"modelName":5,"hardwareId":70,"hardwareName":25,"frameworkId":70,"frameworkName":134,"quantization":262,"hardwareVramGb":26,"decodeTps":428,"prefillTps":429,"vramGb":265,"genSeconds":10,"modelUseCase":8,"ttftS":430,"ttfbGb":431,"mtpAcceptanceRate":10,"powerW":432,"sampleCount":82,"reproduceCount":82,"sourceTypes":433,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":230,"latestOs":110,"latestCudaVersion":143,"latestFlashAttention":132,"latestOutputSpec":10},95.91,524.52,2.051,0.77,283.6,[106],{"modelId":4,"modelName":5,"hardwareId":70,"hardwareName":25,"frameworkId":57,"frameworkName":156,"quantization":113,"hardwareVramGb":26,"decodeTps":435,"prefillTps":436,"vramGb":116,"genSeconds":10,"modelUseCase":8,"ttftS":437,"ttfbGb":438,"mtpAcceptanceRate":10,"powerW":439,"sampleCount":82,"reproduceCount":82,"sourceTypes":440,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":222,"latestOs":110,"latestCudaVersion":111,"latestFlashAttention":93,"latestOutputSpec":10},73.36,312.95,1.551,0.59,292.5,[106],{"modelId":4,"modelName":5,"hardwareId":57,"hardwareName":88,"frameworkId":70,"frameworkName":134,"quantization":157,"hardwareVramGb":17,"decodeTps":442,"prefillTps":443,"vramGb":180,"genSeconds":10,"modelUseCase":8,"ttftS":444,"ttfbGb":82,"mtpAcceptanceRate":10,"powerW":445,"sampleCount":82,"reproduceCount":82,"sourceTypes":446,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":230,"latestOs":110,"latestCudaVersion":111,"latestFlashAttention":93,"latestOutputSpec":10},125.25,439.59,1.299,152.4,[106],{"modelId":4,"modelName":5,"hardwareId":145,"hardwareName":40,"frameworkId":82,"frameworkName":97,"quantization":420,"hardwareVramGb":26,"decodeTps":448,"prefillTps":449,"vramGb":116,"genSeconds":10,"modelUseCase":8,"ttftS":450,"ttfbGb":451,"mtpAcceptanceRate":10,"powerW":452,"sampleCount":82,"reproduceCount":82,"sourceTypes":453,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":214,"latestOs":110,"latestCudaVersion":143,"latestFlashAttention":132,"latestOutputSpec":10},85.47,436.48,2.438,0.68,219.4,[106],{"modelId":4,"modelName":5,"hardwareId":145,"hardwareName":40,"frameworkId":70,"frameworkName":134,"quantization":409,"hardwareVramGb":26,"decodeTps":455,"prefillTps":456,"vramGb":412,"genSeconds":10,"modelUseCase":8,"ttftS":457,"ttfbGb":458,"mtpAcceptanceRate":10,"powerW":459,"sampleCount":82,"reproduceCount":82,"sourceTypes":460,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":230,"latestOs":110,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},102.61,458.68,1.781,0.82,209,[106],{"modelId":4,"modelName":5,"hardwareId":167,"hardwareName":29,"frameworkId":82,"frameworkName":97,"quantization":187,"hardwareVramGb":30,"decodeTps":462,"prefillTps":463,"vramGb":190,"genSeconds":10,"modelUseCase":8,"ttftS":464,"ttfbGb":465,"mtpAcceptanceRate":10,"powerW":466,"sampleCount":82,"reproduceCount":82,"sourceTypes":467,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":468,"latestOs":142,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},313.16,1103.72,1.137,2.51,465.4,[106],"v0.34",{"modelId":4,"modelName":5,"hardwareId":167,"hardwareName":29,"frameworkId":70,"frameworkName":134,"quantization":305,"hardwareVramGb":30,"decodeTps":470,"prefillTps":471,"vramGb":308,"genSeconds":10,"modelUseCase":8,"ttftS":472,"ttfbGb":473,"mtpAcceptanceRate":10,"powerW":474,"sampleCount":82,"reproduceCount":82,"sourceTypes":475,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":303,"latestOs":154,"latestCudaVersion":295,"latestFlashAttention":132,"latestOutputSpec":10},290.25,1286.52,1.617,2.32,403.6,[106],{"modelId":4,"modelName":5,"hardwareId":167,"hardwareName":29,"frameworkId":70,"frameworkName":134,"quantization":286,"hardwareVramGb":30,"decodeTps":477,"prefillTps":478,"vramGb":289,"genSeconds":10,"modelUseCase":8,"ttftS":479,"ttfbGb":480,"mtpAcceptanceRate":10,"powerW":481,"sampleCount":82,"reproduceCount":82,"sourceTypes":482,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":483,"latestOs":154,"latestCudaVersion":295,"latestFlashAttention":132,"latestOutputSpec":10},193.3,1031.83,1.153,1.55,404.2,[106],"v0.32",{"modelId":4,"modelName":5,"hardwareId":7,"hardwareName":76,"frameworkId":82,"frameworkName":97,"quantization":157,"hardwareVramGb":17,"decodeTps":485,"prefillTps":486,"vramGb":180,"genSeconds":10,"modelUseCase":8,"ttftS":487,"ttfbGb":488,"mtpAcceptanceRate":10,"powerW":489,"sampleCount":82,"reproduceCount":82,"sourceTypes":490,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":491,"latestOs":122,"latestCudaVersion":111,"latestFlashAttention":132,"latestOutputSpec":10},202.82,808.1,1.064,1.62,207.9,[106],"v0.41",{"modelId":4,"modelName":5,"hardwareId":17,"hardwareName":69,"frameworkId":82,"frameworkName":97,"quantization":157,"hardwareVramGb":17,"decodeTps":493,"prefillTps":494,"vramGb":180,"genSeconds":10,"modelUseCase":8,"ttftS":495,"ttfbGb":496,"mtpAcceptanceRate":10,"powerW":497,"sampleCount":82,"reproduceCount":82,"sourceTypes":498,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":260,"latestOs":154,"latestCudaVersion":123,"latestFlashAttention":132,"latestOutputSpec":10},150.94,517.81,0.299,1.21,225.4,[106],{"modelId":4,"modelName":5,"hardwareId":186,"hardwareName":74,"frameworkId":27,"frameworkName":500,"quantization":409,"hardwareVramGb":17,"decodeTps":501,"prefillTps":502,"vramGb":180,"genSeconds":10,"modelUseCase":8,"ttftS":503,"ttfbGb":504,"mtpAcceptanceRate":505,"powerW":506,"sampleCount":82,"reproduceCount":82,"sourceTypes":507,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":240,"latestOs":110,"latestCudaVersion":143,"latestFlashAttention":93,"latestOutputSpec":10},"Strata",252.15,1022.42,0.855,2.02,0.7663,240.2,[106],{"modelId":4,"modelName":5,"hardwareId":509,"hardwareName":90,"frameworkId":57,"frameworkName":156,"quantization":409,"hardwareVramGb":17,"decodeTps":510,"prefillTps":511,"vramGb":180,"genSeconds":10,"modelUseCase":8,"ttftS":512,"ttfbGb":414,"mtpAcceptanceRate":10,"powerW":513,"sampleCount":82,"reproduceCount":82,"sourceTypes":514,"evidenceLevel":107,"updatedAt":417,"latestFrameworkVersion":222,"latestOs":154,"latestCudaVersion":295,"latestFlashAttention":132,"latestOutputSpec":10},41,163.57,753.22,0.624,258.5,[106],{"modelId":4,"modelName":5,"hardwareId":17,"hardwareName":69,"frameworkId":57,"frameworkName":156,"quantization":187,"hardwareVramGb":17,"decodeTps":516,"prefillTps":517,"vramGb":190,"genSeconds":10,"modelUseCase":8,"ttftS":518,"ttfbGb":162,"mtpAcceptanceRate":10,"powerW":519,"sampleCount":82,"reproduceCount":82,"sourceTypes":520,"evidenceLevel":107,"updatedAt":521,"latestFrameworkVersion":109,"latestOs":110,"latestCudaVersion":522,"latestFlashAttention":93,"latestOutputSpec":10},91.31,484.29,0.748,311.5,[106],"2026-10-15T13:07:00.777Z","N\u002FA",{"modelId":4,"modelName":5,"hardwareId":524,"hardwareName":84,"frameworkId":70,"frameworkName":134,"quantization":262,"hardwareVramGb":17,"decodeTps":525,"prefillTps":526,"vramGb":180,"genSeconds":10,"modelUseCase":8,"ttftS":527,"ttfbGb":528,"mtpAcceptanceRate":10,"powerW":529,"sampleCount":82,"reproduceCount":82,"sourceTypes":530,"evidenceLevel":107,"updatedAt":521,"latestFrameworkVersion":141,"latestOs":154,"latestCudaVersion":522,"latestFlashAttention":132,"latestOutputSpec":10},21,87.65,404.8,2.073,0.7,297.3,[106],82,{"items":533,"total":82,"page":82,"pageSize":186},[534],{"id":535,"type":536,"title":537,"excerpt":538,"author":539,"replyCount":70,"createdAt":541,"lastRepliedAt":542,"resolvedAt":10,"editedAt":10,"model":543,"hardware":544,"variant":10,"record":10},22,"SHOWCASE","DeepSeek-R1-Distill-Qwen-14B + RTX 4070 12GB + Q5_K_M：R1 推理能力下放到 12G 卡","新闻 #40 把 R1-Distill-14B 推上 12-16GB 卡甜区。我昨晚在 RTX 4070 12GB 上跑了一晚上，汇报。 硬件： RTX 4070 12GB（不是 Ti，普通版） 64GB DDR5 NVMe 4.0 2TB",{"nickname":540,"avatarUrl":10},"matrix_w","2026-10-15T10:30:00.000Z","2026-10-15T12:30:00.000Z",{"name":5},{"name":545},"NVIDIA RTX 4070",{"items":547},[],{"items":549,"total":70,"page":82,"pageSize":57},[550,590],{"slug":551,"category":552,"publishedAt":553,"titleZh":554,"titleEn":555,"summaryZh":556,"summaryEn":557,"models":558,"hardwares":579},"llama-family-kimi-k3-bench-2026-10-18","NEWS","2026-10-18T14:00:00.000Z","Llama 3.x + Kimi-K3 实测：Llama-3.2-3B 跑出 277 tok\u002Fs，Kimi-K3 112 tok\u002Fs 能装 16GB 卡","Llama 3.x + Kimi-K3 benchmarks: Llama-3.2-3B hits 277 tok\u002Fs, Kimi-K3 112 tok\u002Fs fits 16GB cards","10-18 批量首次大规模覆盖 Meta Llama 3.x 全家族 + Kimi-K3：Llama-3.2-3B 平均 276.96、Llama-3.1-8B 171.27、Llama-3-8B 155.78、Kimi-K3 112.33 tok\u002Fs。**Llama-3.2-3B 是 DB 最快 3B 模型**。","10-18 batch first large-sample coverage of Meta Llama 3.x family + Kimi-K3: Llama-3.2-3B avg 276.96, Llama-3.1-8B 171.27, Llama-3-8B 155.78, Kimi-K3 112.33 tok\u002Fs. **Llama-3.2-3B is DB's fastest 3B model**.",[559,562,564,567,569,572,575,576],{"id":560,"name":561},48,"Qwen3.5-4B",{"id":220,"name":563},"Qwen3.5-27B",{"id":565,"name":566},53,"Qwen3-8B",{"id":347,"name":568},"Llama-3.1-8B-Instruct",{"id":570,"name":571},69,"Llama-3-8B-Instruct",{"id":573,"name":574},70,"Llama-3.2-3B-Instruct",{"id":4,"name":5},{"id":577,"name":578},73,"Kimi-K3",[580,581,582,583,584,585,588,589],{"id":82,"name":33},{"id":70,"name":25},{"id":145,"name":40},{"id":167,"name":29},{"id":7,"name":76},{"id":586,"name":587},15,"NVIDIA RTX 3060 12GB",{"id":186,"name":74},{"id":524,"name":84},{"slug":591,"category":552,"publishedAt":592,"titleZh":593,"titleEn":594,"summaryZh":595,"summaryEn":596,"models":597,"hardwares":609},"deepseek-r1-distill-qwen-14b-debut-2026-10-15","2026-10-15T14:00:00.000Z","DeepSeek-R1-Distill-Qwen-14B + Qwen3.5-9B 首次入库：蒸馏模型在 12G 卡跑 130 tok\u002Fs","DeepSeek-R1-Distill-Qwen-14B + Qwen3.5-9B first DB entries: distilled model hits 130 tok\u002Fs on 12GB card","10-13 批量 27 条 Qwen3.5-9B 实测平均 109.44 tok\u002Fs，18 条 DeepSeek-R1-Distill-Qwen-14B 实测平均 108.01 tok\u002Fs。9B 和 14B 蒸馏模型是 12-16G 卡的甜区。","10-13 batch: 27 Qwen3.5-9B records avg 109.44 tok\u002Fs, 18 DeepSeek-R1-Distill-Qwen-14B records avg 108.01 tok\u002Fs. 9B and 14B distilled models are the sweet spot for 12-16GB cards.",[598,601,604,605,608],{"id":599,"name":600},49,"Qwen3.5-9B",{"id":602,"name":603},52,"Qwen3.8-9B",{"id":565,"name":566},{"id":606,"name":607},65,"gemma-4-12B-it",{"id":4,"name":5},[610,611,612,613,616],{"id":57,"name":88},{"id":586,"name":587},{"id":17,"name":69},{"id":614,"name":615},19,"NVIDIA RTX 5060 Ti 16GB",{"id":524,"name":84}]