Download board.json from quantid/home: direct link, hf CLI and curl.
- Browser
- Download file 69.6 kB
-
https://huggingface.co/spaces/quantid/home/resolve/main/board.json
- Command line
-
hf download hf://spaces/quantid/home/board.json
-
curl -L -o board.json https://huggingface.co/spaces/quantid/home/resolve/main/board.json
69.6 kB
| {"updated":"2026-08-27","hw":{"H100 NVL":{"bw":3.9,"mem":94,"kind":"datacenter","w":400,"usd":3.19},"H100 SXM":{"bw":3.35,"mem":80,"kind":"datacenter","w":700,"usd":3.29},"H800":{"bw":3.35,"mem":80,"kind":"datacenter","w":700,"usd":3.29},"H200":{"bw":4.8,"mem":141,"kind":"datacenter","w":null,"usd":4.59},"B200":{"bw":8.0,"mem":180,"kind":"datacenter","w":1000,"usd":6.79},"B300 SXM6":{"bw":8.0,"mem":288,"kind":"datacenter","w":1400,"usd":7.89},"GB200":{"bw":8.0,"mem":186,"kind":"datacenter","w":null,"usd":null},"MI300X":{"bw":5.3,"mem":192,"kind":"datacenter","w":750,"usd":null},"A100 80GB":{"bw":2.039,"mem":80,"kind":"datacenter","w":null,"usd":null},"L40S":{"bw":0.864,"mem":48,"kind":"datacenter","w":350,"usd":0.99},"A10G":{"bw":0.6,"mem":24,"kind":"cloud","w":150,"usd":null},"RTX 5060 Laptop":{"bw":0.448,"mem":8,"kind":"consumer","w":100,"usd":null},"Apple M3 Pro":{"bw":0.15,"mem":18,"kind":"consumer","w":null,"usd":null},"CPU Xeon":{"bw":null,"mem":null,"kind":"cpu","w":null,"usd":null},"CPU (A10G box)":{"bw":null,"mem":null,"kind":"cpu","w":null,"usd":null},"Apple M3 Pro CPU":{"bw":0.15,"mem":18,"kind":"consumer","w":null,"usd":null},"CPU DDR5":{"bw":0.09,"mem":null,"kind":"cpu","w":null,"usd":null},"CPU shared":{"bw":0.02,"mem":16,"kind":"cpu","w":null,"usd":null},"MI355X":{"bw":8.0,"mem":288,"kind":"datacenter","w":1400,"usd":null},"RTX PRO 6000 Blackwell":{"bw":1.79,"mem":96,"kind":"cloud","w":null,"usd":2.09},"B300 SXM":{"bw":8.0,"mem":270,"kind":"datacenter","w":1400,"usd":7.89},"MI350X":{"bw":8.0,"mem":288,"kind":"datacenter","w":null,"usd":null},"H200 NVL":{"bw":4.8,"mem":141,"kind":"datacenter","w":null,"usd":4.59},"GB300":{"bw":8.0,"mem":288,"kind":"datacenter","w":1400,"usd":null},"RTX PRO 4500 Blackwell":{"bw":null,"mem":32,"kind":"consumer","w":null,"usd":null},"Intel Arc Pro B60":{"bw":null,"mem":48,"kind":"consumer","w":null,"usd":null},"MI325X":{"bw":6.0,"mem":256,"kind":"datacenter","w":1000,"usd":null},"H200 SXM":{"bw":4.8,"mem":141,"kind":"datacenter","w":700,"usd":4.59},"GH200 144GB":{"bw":4.9,"mem":144,"kind":"datacenter","w":null,"usd":2.29},"H100 PCIe":{"bw":2.0,"mem":80,"kind":"datacenter","w":350,"usd":2.89},"GH200 96GB":{"bw":4.0,"mem":96,"kind":"datacenter","w":null,"usd":2.29},"Jetson AGX Thor":{"bw":null,"mem":128,"kind":"consumer","w":null,"usd":null}},"bpp":{"FP4":0.5,"NVFP4":0.5,"MXFP4":0.5,"UINT4":0.5,"INT4 (W4A16)":0.5,"FP8":1.0,"INT8":1.0,"FP16":2.0,"BF16":2.0,"FP32":4.0},"records":[{"m":"Llama 3.1 8B","g":"B300 SXM","n":8,"c":null,"t":166745.0,"pg":20843.1,"s":0,"q":"FP4","by":"ASUSTeK","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"B200","n":8,"c":null,"t":160403.0,"pg":20050.4,"s":0,"q":"FP4","by":"Lambda SIT","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"B300 SXM","n":8,"c":null,"t":148067.0,"pg":18508.4,"s":0,"q":"FP4","by":"ASUSTeK","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"B200","n":8,"c":null,"t":146960.0,"pg":18370.0,"s":0,"q":"FP4","by":"Lenovo","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Ourbox-35B","g":"B200","n":1,"c":256,"t":18057.0,"pg":18057.0,"s":70.5,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":["VKIE"],"sw":[[8,1332.0],[32,4231.0],[64,8154.0],[128,12692.0],[256,18057.0],[512,18120.0]]},{"m":"Llama 3.1 8B","g":"B200","n":8,"c":null,"t":142508.0,"pg":17813.5,"s":0,"q":"FP4","by":"Inventec","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"B200","n":8,"c":null,"t":131270.0,"pg":16408.8,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"B200","n":8,"c":null,"t":129047.0,"pg":16130.9,"s":0,"q":"FP4","by":"Supermicro","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"B200","n":8,"c":null,"t":128795.0,"pg":16099.4,"s":0,"q":"FP4","by":"Supermicro","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"B200","n":8,"c":null,"t":128794.0,"pg":16099.2,"s":0,"q":"FP4","by":"Dell","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"B200","n":8,"c":null,"t":128750.0,"pg":16093.8,"s":0,"q":"FP4","by":"Lambda SIT","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"B300 SXM","n":8,"c":null,"t":128633.0,"pg":16079.1,"s":0,"q":"FP4","by":"GigaComputing","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"B200","n":8,"c":null,"t":127209.0,"pg":15901.1,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Ourbox-35B","g":"B200","n":1,"c":512,"t":15816.0,"pg":15816.0,"s":30.9,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":["VKAE","VKIE"]},{"m":"Mixtral 8x7B","g":"B200","n":8,"c":null,"t":126468.0,"pg":15808.5,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GB300","n":72,"c":null,"t":1126850.0,"pg":15650.7,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":1,"c":null,"t":15371.7,"pg":15371.7,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"B200","n":8,"c":null,"t":121886.0,"pg":15235.8,"s":0,"q":"FP4","by":"Lambda","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":72,"c":null,"t":1096770.0,"pg":15232.9,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":4,"c":null,"t":60220.4,"pg":15055.1,"s":0,"q":"FP4","by":"Lambda SIT","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":1,"c":null,"t":14973.1,"pg":14973.1,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":64,"c":null,"t":956523.0,"pg":14945.7,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":72,"c":null,"t":1046150.0,"pg":14529.9,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"B300 SXM","n":8,"c":null,"t":115730.0,"pg":14466.2,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"B300 SXM","n":8,"c":null,"t":114504.0,"pg":14313.0,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":64,"c":null,"t":910468.0,"pg":14226.1,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B300 SXM","n":8,"c":null,"t":112954.0,"pg":14119.2,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B300 SXM","n":8,"c":null,"t":111496.0,"pg":13937.0,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B300 SXM","n":8,"c":null,"t":110655.0,"pg":13831.9,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B300 SXM","n":8,"c":null,"t":108392.0,"pg":13549.0,"s":0,"q":"FP4","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":4,"c":null,"t":53462.7,"pg":13365.7,"s":0,"q":"FP4","by":"Lambda SIT","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":104572.0,"pg":13071.5,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"B200","n":4,"c":null,"t":52061.6,"pg":13015.4,"s":0,"q":"FP4","by":"Azure","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"B200","n":8,"c":null,"t":103954.0,"pg":12994.2,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"B200","n":8,"c":null,"t":103635.0,"pg":12954.4,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":103627.0,"pg":12953.4,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":8,"c":null,"t":103480.0,"pg":12935.0,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":4,"c":null,"t":51736.9,"pg":12934.2,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":102909.0,"pg":12863.6,"s":0,"q":"FP4","by":"Lenovo","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":101611.0,"pg":12701.4,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"B200","n":8,"c":null,"t":100291.0,"pg":12536.4,"s":0,"q":"FP4","by":"Inventec","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":8,"c":null,"t":100282.0,"pg":12535.2,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":72,"c":null,"t":899218.0,"pg":12489.1,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":98858.0,"pg":12357.2,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":4,"c":null,"t":49359.6,"pg":12339.9,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":72,"c":null,"t":888054.0,"pg":12334.1,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":98443.3,"pg":12305.4,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":1,"c":null,"t":12269.3,"pg":12269.3,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":72,"c":null,"t":879542.0,"pg":12215.9,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":97403.4,"pg":12175.4,"s":0,"q":"FP4","by":"Lambda","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":64,"c":null,"t":774539.0,"pg":12102.2,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":1,"c":null,"t":12078.4,"pg":12078.4,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GB300","n":72,"c":null,"t":868278.0,"pg":12059.4,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"MI355X","n":8,"c":null,"t":95004.0,"pg":11875.5,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":88,"c":null,"t":1042110.0,"pg":11842.2,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Ourbox-35B","g":"B200","n":2,"c":512,"t":23292.0,"pg":11646.0,"s":493.0,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":["VKIE"]},{"m":"gpt-oss-120b","g":"B200","n":8,"c":null,"t":93070.7,"pg":11633.8,"s":0,"q":"MXFP4","by":"RedHat","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":8,"c":null,"t":93045.8,"pg":11630.7,"s":0,"q":"MXFP4","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":64,"c":null,"t":743550.0,"pg":11618.0,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":88,"c":null,"t":1016380.0,"pg":11549.8,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GB300","n":72,"c":null,"t":814128.0,"pg":11307.3,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":72,"c":null,"t":810104.0,"pg":11251.4,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":32,"c":null,"t":350820.0,"pg":10963.1,"s":0,"q":"MXFP4","by":"AMD MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":8,"c":null,"t":87444.2,"pg":10930.5,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"MI355X","n":96,"c":null,"t":1031070.0,"pg":10740.3,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":72,"c":null,"t":754855.0,"pg":10484.1,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"MI355X","n":8,"c":null,"t":82136.1,"pg":10267.0,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":64,"c":null,"t":648248.0,"pg":10128.9,"s":0,"q":"MXFP4","by":"AMD MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"MI350X","n":8,"c":null,"t":80478.1,"pg":10059.8,"s":0,"q":"FP4","by":"Supermicro RedHat","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"MI350X","n":16,"c":null,"t":159932.0,"pg":9995.8,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI350X","n":8,"c":null,"t":78988.6,"pg":9873.6,"s":0,"q":"FP4","by":"MiTAC","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":8,"c":null,"t":78564.0,"pg":9820.5,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":64,"c":null,"t":627653.0,"pg":9807.1,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI350X","n":8,"c":null,"t":77377.9,"pg":9672.2,"s":0,"q":"FP4","by":"MiTAC","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"MI350X","n":16,"c":null,"t":153022.0,"pg":9563.9,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"GB300","n":72,"c":null,"t":677199.0,"pg":9405.5,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"MI355X","n":96,"c":null,"t":900054.0,"pg":9375.6,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":72,"c":null,"t":673936.0,"pg":9360.2,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":8,"c":null,"t":73608.2,"pg":9201.0,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 4 Maverick","g":"B200","n":8,"c":null,"t":72000.0,"pg":9000.0,"s":1000.0,"q":"FP8","by":"NVIDIA","gr":"REPORTED","ab":17,"tb":400,"ctx":0,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":88,"c":null,"t":785522.0,"pg":8926.4,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B300 SXM","n":8,"c":null,"t":70724.3,"pg":8840.5,"s":0,"q":"FP4","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B300 SXM","n":8,"c":null,"t":70325.7,"pg":8790.7,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"MI325X","n":8,"c":null,"t":70035.6,"pg":8754.5,"s":0,"q":"FP8","by":"Vultr","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":72,"c":null,"t":624929.0,"pg":8679.6,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":288,"c":null,"t":2494310.0,"pg":8660.8,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"GH200 144GB","n":2,"c":null,"t":16703.4,"pg":8351.7,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"GH200 144GB","n":1,"c":null,"t":8345.7,"pg":8345.7,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 SXM","n":8,"c":null,"t":66036.8,"pg":8254.6,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":65770.1,"pg":8221.3,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 SXM","n":8,"c":null,"t":64914.6,"pg":8114.3,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":8,"c":null,"t":64510.0,"pg":8063.8,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"GH200 144GB","n":1,"c":null,"t":8042.3,"pg":8042.3,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"MI350X","n":8,"c":null,"t":64293.3,"pg":8036.7,"s":0,"q":"FP4","by":"Supermicro RedHat","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":72,"c":null,"t":575580.0,"pg":7994.2,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 SXM","n":8,"c":null,"t":63514.7,"pg":7939.3,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 SXM","n":4,"c":null,"t":31503.7,"pg":7875.9,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":62851.2,"pg":7856.4,"s":0,"q":"FP4","by":"GigaComputing","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":62624.9,"pg":7828.1,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"GH200 96GB","n":1,"c":null,"t":7792.9,"pg":7792.9,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"GH200 144GB","n":2,"c":null,"t":15570.8,"pg":7785.4,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"GH200 96GB","n":1,"c":null,"t":7771.8,"pg":7771.8,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 SXM","n":8,"c":null,"t":62108.4,"pg":7763.6,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 SXM","n":8,"c":null,"t":61802.0,"pg":7725.2,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"B200","n":8,"c":null,"t":61644.7,"pg":7705.6,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":8,"c":null,"t":61300.4,"pg":7662.6,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 SXM","n":8,"c":null,"t":60955.1,"pg":7619.4,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B300 SXM","n":8,"c":null,"t":60413.4,"pg":7551.7,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"MI325X","n":8,"c":null,"t":59783.6,"pg":7472.9,"s":0,"q":"FP8","by":"AMD","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"B200","n":4,"c":null,"t":29745.6,"pg":7436.4,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 SXM","n":4,"c":null,"t":29301.5,"pg":7325.4,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":64,"c":null,"t":468665.0,"pg":7322.9,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":8,"c":null,"t":58581.6,"pg":7322.7,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"MI350X","n":8,"c":null,"t":58373.3,"pg":7296.7,"s":0,"q":"FP4","by":"Supermicro RedHat","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI350X","n":8,"c":null,"t":54730.5,"pg":6841.3,"s":0,"q":"FP4","by":"MiTAC","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":72,"c":null,"t":486141.0,"pg":6752.0,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"MI300X","n":8,"c":null,"t":53333.9,"pg":6666.7,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 SXM","n":8,"c":null,"t":53299.3,"pg":6662.4,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":8,"c":null,"t":53193.6,"pg":6649.2,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 SXM","n":8,"c":null,"t":52724.4,"pg":6590.6,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 NVL","n":8,"c":null,"t":52205.9,"pg":6525.7,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"MI350X","n":16,"c":null,"t":103633.0,"pg":6477.1,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":8,"c":null,"t":51692.9,"pg":6461.6,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":4,"c":null,"t":25753.8,"pg":6438.4,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":4,"c":null,"t":25731.6,"pg":6432.9,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":8,"c":null,"t":50776.8,"pg":6347.1,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":24,"c":null,"t":151843.0,"pg":6326.8,"s":0,"q":"FP8","by":"Dell MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":4,"c":null,"t":25180.8,"pg":6295.2,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H200 NVL","n":8,"c":null,"t":50123.9,"pg":6265.5,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":12495.2,"pg":6247.6,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":49580.2,"pg":6197.5,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":48794.3,"pg":6099.3,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":60679.9,"pg":6068.0,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":4,"c":null,"t":24216.6,"pg":6054.1,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":8,"c":null,"t":48047.0,"pg":6005.9,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":24013.5,"pg":6003.4,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"MI300X","n":8,"c":null,"t":47804.3,"pg":5975.5,"s":0,"q":"FP8","by":"AMD","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":59632.7,"pg":5963.3,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":11899.7,"pg":5949.9,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI355X","n":24,"c":null,"t":141521.0,"pg":5896.7,"s":0,"q":"FP8","by":"Dell MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":23423.9,"pg":5856.0,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":46841.8,"pg":5855.2,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":72,"c":null,"t":420659.0,"pg":5842.5,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H100 SXM","n":1,"c":null,"t":5777.1,"pg":5777.1,"s":0,"q":"FP8","by":"RedHat","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":46060.0,"pg":5757.5,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":8,"c":null,"t":45075.8,"pg":5634.5,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":44087.3,"pg":5510.9,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":288,"c":null,"t":1555110.0,"pg":5399.7,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":53054.7,"pg":5305.5,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":10589.3,"pg":5294.6,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H100 SXM","n":1,"c":null,"t":5104.0,"pg":5104.0,"s":0,"q":"FP8","by":"RedHat","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":20118.9,"pg":5029.7,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 NVL","n":4,"c":null,"t":19500.9,"pg":4875.2,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":64,"c":null,"t":311531.0,"pg":4867.7,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"H200 NVL","n":8,"c":null,"t":38696.9,"pg":4837.1,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 NVL","n":4,"c":null,"t":19334.9,"pg":4833.7,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 NVL","n":8,"c":null,"t":37934.5,"pg":4741.8,"s":0,"q":"FP8","by":"Fujitsu","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 NVL","n":8,"c":null,"t":37713.6,"pg":4714.2,"s":0,"q":"FP8","by":"Fujitsu","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 144GB","n":1,"c":null,"t":4686.1,"pg":4686.1,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 144GB","n":2,"c":null,"t":9362.9,"pg":4681.4,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":72,"c":null,"t":336106.0,"pg":4668.1,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":35453.0,"pg":4431.6,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":35316.7,"pg":4414.6,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":4,"c":null,"t":17654.1,"pg":4413.5,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 144GB","n":1,"c":null,"t":4350.3,"pg":4350.3,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 144GB","n":2,"c":null,"t":8674.6,"pg":4337.3,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":8,"c":null,"t":34555.0,"pg":4319.4,"s":0,"q":"FP8","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":16,"c":null,"t":68894.8,"pg":4305.9,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"H200 SXM","n":8,"c":null,"t":34193.8,"pg":4274.2,"s":0,"q":"FP8","by":"ASUSTeK","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":34029.4,"pg":4253.7,"s":0,"q":"FP8","by":"Nebius","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":8,"c":null,"t":33928.1,"pg":4241.0,"s":0,"q":"FP8","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":8,"c":null,"t":33880.4,"pg":4235.1,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":33637.3,"pg":4204.7,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":8,"c":null,"t":33379.0,"pg":4172.4,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":16,"c":null,"t":66439.0,"pg":4152.4,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":33071.5,"pg":4133.9,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":41242.4,"pg":4124.2,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 144GB","n":2,"c":null,"t":8231.9,"pg":4115.9,"s":0,"q":"FP8","by":"Supermicro","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":8198.2,"pg":4099.1,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":32738.8,"pg":4092.3,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":16,"c":null,"t":65320.1,"pg":4082.5,"s":0,"q":"FP8","by":"Supermicro MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":72,"c":null,"t":289712.0,"pg":4023.8,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":32151.0,"pg":4018.9,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":8,"c":null,"t":32139.4,"pg":4017.4,"s":0,"q":"FP8","by":"GigaComputing","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":32004.0,"pg":4000.5,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":8,"c":null,"t":31888.9,"pg":3986.1,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 PCIe","n":4,"c":null,"t":15843.3,"pg":3960.8,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":39234.1,"pg":3923.4,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":31383.2,"pg":3922.9,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":8,"c":null,"t":31306.8,"pg":3913.3,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":31267.4,"pg":3908.4,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":32,"c":null,"t":124879.0,"pg":3902.5,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":16,"c":null,"t":62351.6,"pg":3897.0,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":31149.9,"pg":3893.7,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":8,"c":null,"t":31106.3,"pg":3888.3,"s":0,"q":"FP8","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":7742.8,"pg":3871.4,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":8,"c":null,"t":30724.5,"pg":3840.6,"s":0,"q":"FP8","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":32,"c":null,"t":122274.0,"pg":3821.1,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":16,"c":null,"t":61101.3,"pg":3818.8,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":4,"c":null,"t":15167.1,"pg":3791.8,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":30304.0,"pg":3788.0,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 144GB","n":2,"c":null,"t":7538.8,"pg":3769.4,"s":0,"q":"FP8","by":"Supermicro","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":30134.3,"pg":3766.8,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"H200 NVL","n":4,"c":null,"t":15057.6,"pg":3764.4,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":30042.8,"pg":3755.3,"s":0,"q":"FP16","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":29908.4,"pg":3738.6,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":29900.1,"pg":3737.5,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":14828.3,"pg":3707.1,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":36710.8,"pg":3671.1,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":36704.2,"pg":3670.4,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":16,"c":null,"t":58617.3,"pg":3663.6,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":4,"c":null,"t":14652.8,"pg":3663.2,"s":0,"q":"FP8","by":"Lenovo","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":24,"c":null,"t":87334.0,"pg":3638.9,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":29084.9,"pg":3635.6,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":29070.3,"pg":3633.8,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":24,"c":null,"t":86558.8,"pg":3606.6,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":28848.4,"pg":3606.1,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":28727.8,"pg":3591.0,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":28717.1,"pg":3589.6,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":16,"c":null,"t":57391.8,"pg":3587.0,"s":0,"q":"FP8","by":"Supermicro MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"H200 SXM","n":8,"c":null,"t":28680.0,"pg":3585.0,"s":0,"q":"MXFP4","by":"RedHat","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":35590.8,"pg":3559.1,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":28421.0,"pg":3552.6,"s":0,"q":"FP16","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":35522.3,"pg":3552.2,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":16,"c":null,"t":56544.8,"pg":3534.1,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":27854.4,"pg":3481.8,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":72,"c":null,"t":250634.0,"pg":3481.0,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":6953.5,"pg":3476.8,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":6953.0,"pg":3476.5,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"H100 PCIe","n":4,"c":null,"t":13864.5,"pg":3466.1,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Mixtral 8x7B","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":13846.1,"pg":3461.5,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":12.9,"tb":46.7,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":6901.9,"pg":3450.9,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":27598.7,"pg":3449.8,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":2,"c":null,"t":6893.4,"pg":3446.7,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"H200 NVL","n":4,"c":null,"t":13736.1,"pg":3434.0,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":4,"c":null,"t":13730.7,"pg":3432.7,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":27185.5,"pg":3398.2,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":13529.8,"pg":3382.4,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":16,"c":null,"t":53572.2,"pg":3348.3,"s":0,"q":"FP8","by":"Dell MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":72,"c":null,"t":240318.0,"pg":3337.8,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":13200.3,"pg":3300.1,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":26205.8,"pg":3275.7,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":26205.3,"pg":3275.7,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":13029.4,"pg":3257.3,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":26005.9,"pg":3250.7,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B300 SXM","n":8,"c":null,"t":26005.8,"pg":3250.7,"s":0,"q":"FP4","by":"ASUSTeK","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":26001.0,"pg":3250.1,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":25939.8,"pg":3242.5,"s":0,"q":"FP16","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":32,"c":null,"t":103182.0,"pg":3224.4,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 96GB","n":1,"c":null,"t":3203.9,"pg":3203.9,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":25463.1,"pg":3182.9,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"GH200 96GB","n":1,"c":null,"t":3144.2,"pg":3144.2,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":12572.6,"pg":3143.2,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":24860.5,"pg":3107.6,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":24821.6,"pg":3102.7,"s":0,"q":"FP16","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"H200 SXM","n":8,"c":null,"t":24103.2,"pg":3012.9,"s":0,"q":"MXFP4","by":"RedHat","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":16,"c":null,"t":47561.7,"pg":2972.6,"s":0,"q":"FP8","by":"Dell MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":32,"c":null,"t":93039.9,"pg":2907.5,"s":0,"q":"FP8","by":"MangoBoost","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":72,"c":null,"t":209328.0,"pg":2907.3,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 SXM","n":8,"c":null,"t":23079.9,"pg":2885.0,"s":0,"q":"FP8","by":"Nebius","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":8,"c":null,"t":22819.6,"pg":2852.4,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"GB300","n":8,"c":null,"t":22545.1,"pg":2818.1,"s":0,"q":"FP4","by":"CoreWeave","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":8,"c":null,"t":21433.3,"pg":2679.2,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI325X","n":8,"c":null,"t":18846.1,"pg":2355.8,"s":0,"q":"FP8","by":"MiTAC","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 4500 Blackwell","n":1,"c":null,"t":2340.8,"pg":2340.8,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 4500 Blackwell","n":5,"c":null,"t":11665.0,"pg":2333.0,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":72,"c":null,"t":167578.0,"pg":2327.5,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B200","n":8,"c":null,"t":18592.2,"pg":2324.0,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":18561.4,"pg":2320.2,"s":0,"q":"FP16","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Qwen3.8-27B","g":"H100 NVL","n":1,"c":128,"t":2302.9,"pg":2302.9,"s":152.1,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":27,"tb":27,"ctx":32768,"own":1,"st":["VKIE"]},{"m":"Ourbox-35B","g":"B200","n":4,"c":512,"t":9179.0,"pg":2294.8,"s":17.9,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":["VKIE"]},{"m":"Llama 3.1 8B","g":"RTX PRO 4500 Blackwell","n":5,"c":null,"t":11352.1,"pg":2270.4,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 NVL","n":8,"c":null,"t":18084.9,"pg":2260.6,"s":0,"q":"FP8","by":"Fujitsu","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 NVL","n":8,"c":null,"t":16848.6,"pg":2106.1,"s":0,"q":"FP8","by":"Fujitsu","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"RTX PRO 4500 Blackwell","n":1,"c":null,"t":2081.3,"pg":2081.3,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":16419.2,"pg":2052.4,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":8,"c":null,"t":16343.8,"pg":2043.0,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":8,"c":null,"t":16167.4,"pg":2020.9,"s":0,"q":"FP4","by":"Lambda SIT","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":15863.7,"pg":1983.0,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"H100 NVL","n":2,"c":null,"t":3879.9,"pg":1940.0,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":19040.8,"pg":1904.1,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":15189.9,"pg":1898.7,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-V3 / R1","g":"H800","n":8,"c":null,"t":14800.0,"pg":1850.0,"s":21.0,"q":"FP8","by":"DeepSeek","gr":"REPORTED","ab":37,"tb":671,"ctx":0,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":7297.1,"pg":1824.3,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":14258.9,"pg":1782.4,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"RTX PRO 6000 Blackwell","n":10,"c":null,"t":17737.6,"pg":1773.8,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 PCIe","n":4,"c":null,"t":7013.2,"pg":1753.3,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":13916.2,"pg":1739.5,"s":0,"q":"FP16","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"Intel Arc Pro B60","n":4,"c":null,"t":6762.3,"pg":1690.6,"s":0,"q":"INT8","by":"Intel","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H200 NVL","n":4,"c":null,"t":6713.0,"pg":1678.2,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"H200 NVL","n":4,"c":null,"t":6709.7,"pg":1677.4,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"RTX PRO 6000 Blackwell","n":4,"c":null,"t":6687.2,"pg":1671.8,"s":0,"q":"FP4","by":"HPE","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"H100 NVL","n":2,"c":null,"t":3337.8,"pg":1668.9,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"B200","n":8,"c":null,"t":13155.3,"pg":1644.4,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"L40S","n":1,"c":null,"t":1642.2,"pg":1642.2,"s":0,"q":"FP8","by":"RedHat","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"Intel Arc Pro B60","n":4,"c":null,"t":6472.4,"pg":1618.1,"s":0,"q":"UINT4","by":"Intel","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"Intel Arc Pro B60","n":4,"c":null,"t":6293.9,"pg":1573.5,"s":0,"q":"INT8","by":"Intel","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"Intel Arc Pro B60","n":4,"c":null,"t":5348.4,"pg":1337.1,"s":0,"q":"UINT4","by":"Intel","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"H100 SXM","n":8,"c":null,"t":10180.1,"pg":1272.5,"s":0,"q":"UNSET","by":"Krai","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Solar-Open2-250B","g":"H100 NVL","n":2,"c":64,"t":2464.9,"pg":1232.5,"s":102.7,"q":"INT4 (W4A16)","by":"VIDRAFT","gr":"VERIFIED","ab":7.9,"tb":250,"ctx":16384,"own":1,"st":["VKIE"]},{"m":"Llama 2 70B (99.9%)","g":"H100 PCIe","n":4,"c":null,"t":4889.9,"pg":1222.5,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 3.1 8B","g":"L40S","n":1,"c":null,"t":1207.1,"pg":1207.1,"s":0,"q":"FP8","by":"RedHat","gr":"GOLD","ab":8,"tb":8,"ctx":null,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"H200 SXM","n":8,"c":null,"t":8904.8,"pg":1113.1,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"MI300X","n":8,"c":null,"t":8840.4,"pg":1105.0,"s":0,"q":"FP8","by":"AMD","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"Intel Arc Pro B60","n":4,"c":null,"t":3270.7,"pg":817.7,"s":0,"q":"INT8","by":"Intel","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99%)","g":"RTX PRO 6000 Blackwell","n":8,"c":null,"t":6262.6,"pg":782.8,"s":0,"q":"FP4","by":"ASUSTeK","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"Intel Arc Pro B60","n":4,"c":null,"t":3009.4,"pg":752.4,"s":0,"q":"UINT4","by":"Intel","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"DeepSeek-R1","g":"B300 SXM","n":8,"c":null,"t":4934.6,"pg":616.8,"s":0,"q":"FP4","by":"GigaComputing","gr":"GOLD","ab":37,"tb":671,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"Jetson AGX Thor","n":1,"c":null,"t":573.8,"pg":573.8,"s":0,"q":"FP16","by":"NVIDIA","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"Intel Arc Pro B60","n":4,"c":null,"t":2199.5,"pg":549.9,"s":0,"q":"INT8","by":"Intel","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"Intel Arc Pro B60","n":4,"c":null,"t":2110.1,"pg":527.5,"s":0,"q":"UINT4","by":"Intel","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"gemma-4-E4B-it","g":"A10G","n":1,"c":1,"t":510.6,"pg":510.6,"s":510.58,"q":"QAT","by":"VIDRAFT","gr":"GOLD","ab":4,"tb":4,"ctx":4096,"own":1,"st":["VKAE"]},{"m":"Ourbox-35B","g":"A100 80GB","n":1,"c":96,"t":462.9,"pg":462.9,"s":4.8,"q":"Q4_K_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":8192,"own":1,"st":["VKIE"],"sw":[[1,127.0],[8,287.8],[48,456.8],[96,462.9],[128,448.8],[256,399.0]]},{"m":"Llama 2 70B (99.9%)","g":"L40S","n":8,"c":null,"t":3655.9,"pg":457.0,"s":0,"q":"FP16","by":"HPE","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Ourbox-35B","g":"B200","n":1,"c":1,"t":455.0,"pg":455.0,"s":455.0,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":["VKAE"]},{"m":"Ourbox-35B","g":"B200","n":1,"c":1,"t":445.6,"pg":445.6,"s":445.65,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":["VKAE"],"sw":[[1,445.6],[32,5312.0],[512,15816.0]]},{"m":"gpt-oss-120b","g":"Intel Arc Pro B60","n":4,"c":null,"t":1601.9,"pg":400.5,"s":0,"q":"MXFP4","by":"Intel","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Llama 2 70B (99.9%)","g":"L40S","n":8,"c":null,"t":3201.1,"pg":400.1,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":70,"tb":70,"ctx":2048,"own":0,"st":[]},{"m":"Solar-Open2-250B","g":"B300 SXM6","n":4,"c":32,"t":1468.2,"pg":367.1,"s":144.5,"q":"NVFP4 (W4A4)","by":"Nota AI","gr":"REPORTED","ab":7.9,"tb":250,"ctx":32768,"own":0,"st":[]},{"m":"Darwin-36B-Opus","g":"B200","n":1,"c":1,"t":280.8,"pg":280.8,"s":280.8,"q":"BF16","by":"VIDRAFT","gr":"VERIFIED","ab":null,"tb":36,"ctx":32768,"own":1,"st":["VKAE"]},{"m":"Llama 3.1 405B","g":"GB300","n":72,"c":null,"t":19511.7,"pg":271.0,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Qwen3-8B","g":"B200","n":1,"c":1,"t":262.6,"pg":262.6,"s":262.6,"q":"BF16","by":"VIDRAFT","gr":"VERIFIED","ab":8,"tb":8,"ctx":8192,"own":1,"st":["VKAE"],"sw":[[1,262.6],[32,7801.0]]},{"m":"Llama 3.1 405B","g":"GB300","n":72,"c":null,"t":18627.6,"pg":258.7,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"GB300","n":72,"c":null,"t":18365.4,"pg":255.1,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"MI355X","n":8,"c":null,"t":2019.0,"pg":252.4,"s":0,"q":"MXFP4","by":"AMD","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"MI355X","n":8,"c":null,"t":2011.2,"pg":251.4,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B300 SXM","n":8,"c":null,"t":1971.1,"pg":246.4,"s":0,"q":"FP4","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Qwen3.5-9B","g":"B200","n":1,"c":1,"t":237.7,"pg":237.7,"s":237.7,"q":"BF16","by":"VIDRAFT","gr":"VERIFIED","ab":9,"tb":9,"ctx":8192,"own":1,"st":["VKAE"],"sw":[[1,237.7],[32,5180.0]]},{"m":"Llama 3.1 405B","g":"GB300","n":72,"c":null,"t":16104.1,"pg":223.7,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"gpt-oss-120b","g":"Intel Arc Pro B60","n":4,"c":null,"t":884.2,"pg":221.1,"s":0,"q":"MXFP4","by":"Intel","gr":"GOLD","ab":5.1,"tb":117,"ctx":null,"own":0,"st":[]},{"m":"Ourbox-35B","g":"B200","n":1,"c":1,"t":220.8,"pg":220.8,"s":220.8,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":["VKAE"]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":15461.5,"pg":214.7,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":4,"c":null,"t":855.8,"pg":213.9,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":1660.7,"pg":207.6,"s":0,"q":"FP4","by":"Dell","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":14774.3,"pg":205.2,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":1632.1,"pg":204.0,"s":0,"q":"FP4","by":"Inventec","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":4,"c":null,"t":801.9,"pg":200.5,"s":0,"q":"FP4","by":"coreweave","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Ourbox-35B (POCKET)","g":"H100 NVL","n":1,"c":1,"t":197.0,"pg":197.0,"s":196.98,"q":"IQ1_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":4096,"own":1,"st":["POCKET"]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":14133.9,"pg":196.3,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":14010.1,"pg":194.6,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":13886.4,"pg":192.9,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":1538.2,"pg":192.3,"s":0,"q":"FP4","by":"Supermicro","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B300 SXM","n":8,"c":null,"t":1484.4,"pg":185.6,"s":0,"q":"FP4","by":"Quanta Cloud Technology","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"GB300","n":72,"c":null,"t":12248.4,"pg":170.1,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":1350.2,"pg":168.8,"s":0,"q":"FP4","by":"Google","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":11614.3,"pg":161.3,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":1279.5,"pg":159.9,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":4,"c":null,"t":596.1,"pg":149.0,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":9921.0,"pg":137.8,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":9897.5,"pg":137.5,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":1080.3,"pg":135.0,"s":0,"q":"FP4","by":"Supermicro","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Ourbox-35B","g":"A100 80GB","n":1,"c":1,"t":131.9,"pg":131.9,"s":131.9,"q":"Q4_K_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":8192,"own":1,"st":["VKAE"]},{"m":"Llama 3.1 405B","g":"B200","n":4,"c":null,"t":522.1,"pg":130.5,"s":0,"q":"FP4","by":"coreweave","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Ourbox-35B","g":"A10G","n":1,"c":1,"t":126.0,"pg":126.0,"s":126.0,"q":"Q3_K_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":8192,"own":1,"st":["VKUE"]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":983.6,"pg":123.0,"s":0,"q":"FP4","by":"Google","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":72,"c":null,"t":8850.1,"pg":122.9,"s":0,"q":"FP4","by":"NVIDIA","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"MI355X","n":8,"c":null,"t":794.0,"pg":99.2,"s":0,"q":"FP4","by":"AMD","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B200","n":8,"c":null,"t":770.7,"pg":96.3,"s":0,"q":"FP4","by":"Nebius","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"B300 SXM","n":8,"c":null,"t":764.7,"pg":95.6,"s":0,"q":"FP4","by":"GigaComputing","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Qwen3.6-27B","g":"B200","n":1,"c":1,"t":91.0,"pg":91.0,"s":91.0,"q":"BF16","by":"VIDRAFT","gr":"VERIFIED","ab":27,"tb":27,"ctx":8192,"own":1,"st":["VKAE"],"sw":[[1,91.0],[32,2156.0]]},{"m":"Bonsai-27B","g":"H100 NVL","n":1,"c":1,"t":88.7,"pg":88.7,"s":88.72,"q":"Q1_0 (GGUF)","by":"prism-ml","gr":"VERIFIED","ab":27,"tb":27,"ctx":4096,"own":1,"st":[]},{"m":"JGOS-31B-Citizen","g":"B200","n":1,"c":1,"t":81.2,"pg":81.2,"s":81.2,"q":"BF16","by":"VIDRAFT","gr":"REPORTED","ab":31,"tb":31,"ctx":null,"own":1,"st":["VKAE"]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":8,"c":null,"t":573.6,"pg":71.7,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":8,"c":null,"t":564.7,"pg":70.6,"s":0,"q":"FP8","by":"Supermicro","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":8,"c":null,"t":552.7,"pg":69.1,"s":0,"q":"FP8","by":"Nebius","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":16,"c":null,"t":1086.0,"pg":67.9,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"JGOS-398B","g":"B200","n":6,"c":1,"t":382.0,"pg":63.7,"s":382.0,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":null,"tb":398,"ctx":32768,"own":1,"st":["VKAE"]},{"m":"Llama 3.1 405B","g":"H100 SXM","n":8,"c":null,"t":436.2,"pg":54.5,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H200 NVL","n":8,"c":null,"t":428.9,"pg":53.6,"s":0,"q":"FP8","by":"HPE","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H100 SXM","n":16,"c":null,"t":783.1,"pg":48.9,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H100 SXM","n":32,"c":null,"t":1561.6,"pg":48.8,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Armoring-Qwen3.5-9B","g":"RTX 5060 Laptop","n":1,"c":1,"t":42.8,"pg":42.8,"s":42.8,"q":"Q4_0 (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":9.2,"tb":9.2,"ctx":32768,"own":1,"st":["ARMORING"]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":8,"c":null,"t":295.5,"pg":36.9,"s":0,"q":"FP8","by":"Quanta Cloud Technology","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":8,"c":null,"t":294.3,"pg":36.8,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":16,"c":null,"t":584.3,"pg":36.5,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H100 SXM","n":32,"c":null,"t":1126.9,"pg":35.2,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H100 SXM","n":16,"c":null,"t":563.4,"pg":35.2,"s":0,"q":"FP8","by":"Cisco","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"MI300X","n":8,"c":null,"t":278.6,"pg":34.8,"s":0,"q":"FP16","by":"Krai","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Llama 3.1 405B","g":"H100 SXM","n":8,"c":null,"t":277.1,"pg":34.6,"s":0,"q":"FP8","by":"Dell","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"Ourbox-35B (POCKET)","g":"CPU Xeon","n":1,"c":1,"t":27.0,"pg":27.0,"s":27.04,"q":"IQ1_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":4096,"own":1,"st":["POCKET"]},{"m":"Ourbox-35B","g":"B200","n":1,"c":1,"t":25.7,"pg":25.7,"s":25.7,"q":"FP8","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":32768,"own":1,"st":[]},{"m":"Ourbox-35B (POCKET)","g":"Apple M3 Pro","n":1,"c":1,"t":25.4,"pg":25.4,"s":25.4,"q":"IQ1_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":4096,"own":1,"st":["POCKET"]},{"m":"Llama 3.1 405B","g":"H200 SXM","n":8,"c":null,"t":202.8,"pg":25.4,"s":0,"q":"FP8","by":"Nebius","gr":"GOLD","ab":405,"tb":405,"ctx":null,"own":0,"st":[]},{"m":"JGOS-31B-Citizen","g":"L40S","n":1,"c":1,"t":21.0,"pg":21.0,"s":21.0,"q":"INT4 (W4A16 g128)","by":"VIDRAFT","gr":"REPORTED","ab":31,"tb":31,"ctx":null,"own":1,"st":["VKAE"]},{"m":"Ourbox-35B","g":"CPU (A10G box)","n":1,"c":1,"t":20.9,"pg":20.9,"s":20.9,"q":"Q3_K_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":8192,"own":1,"st":["VKUE"]},{"m":"Ourbox-35B","g":"RTX 5060 Laptop","n":1,"c":1,"t":20.0,"pg":20.0,"s":20.01,"q":"Q3_K_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":8192,"own":1,"st":["VKUE"]},{"m":"Ourbox-35B (POCKET)","g":"CPU DDR5","n":1,"c":1,"t":20.0,"pg":20.0,"s":20.0,"q":"Q2_K (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":8192,"own":1,"st":["POCKET"]},{"m":"Ourbox-35B (POCKET)","g":"Apple M3 Pro CPU","n":1,"c":1,"t":19.5,"pg":19.5,"s":19.5,"q":"Q2_K (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":4096,"own":1,"st":["POCKET"]},{"m":"Ourbox-35B (POCKET)","g":"CPU DDR5","n":1,"c":1,"t":17.0,"pg":17.0,"s":17.0,"q":"Q4_K_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":8192,"own":1,"st":["POCKET"]},{"m":"Ourbox-35B (POCKET)","g":"Apple M3 Pro CPU","n":1,"c":1,"t":13.8,"pg":13.8,"s":13.8,"q":"IQ1_M (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":4096,"own":1,"st":["POCKET"]},{"m":"Bonsai-27B","g":"Apple M3 Pro","n":1,"c":1,"t":12.8,"pg":12.8,"s":12.8,"q":"Q1_0 (GGUF)","by":"prism-ml","gr":"VERIFIED","ab":27,"tb":27,"ctx":4096,"own":1,"st":[]},{"m":"Bonsai-27B","g":"CPU Xeon","n":1,"c":1,"t":10.1,"pg":10.1,"s":10.05,"q":"Q1_0 (GGUF)","by":"prism-ml","gr":"VERIFIED","ab":27,"tb":27,"ctx":4096,"own":1,"st":[]},{"m":"Ourbox-35B (POCKET)","g":"CPU shared","n":1,"c":1,"t":6.0,"pg":6.0,"s":6.0,"q":"IQ2 (GGUF)","by":"VIDRAFT","gr":"VERIFIED","ab":3.0,"tb":34.7,"ctx":4096,"own":1,"st":["POCKET","VKUE"]},{"m":"Qwen2.5-32B","g":"RTX 5060 Laptop","n":1,"c":1,"t":5.4,"pg":5.4,"s":5.36,"q":"Q3_K_M (GGUF)","by":"Qwen","gr":"VERIFIED","ab":32.8,"tb":32.8,"ctx":8192,"own":1,"st":[]},{"m":"Bonsai-27B","g":"Apple M3 Pro CPU","n":1,"c":1,"t":4.4,"pg":4.4,"s":4.4,"q":"Q1_0 (GGUF)","by":"prism-ml","gr":"VERIFIED","ab":27,"tb":27,"ctx":4096,"own":1,"st":[]}],"src":{"power":"rated TDP","power_note":"Rated board power from the vendor datasheet, not measured wall power. Real draw during decode is usually below TDP, and a system also spends power on CPUs, memory, networking and cooling that this does not count. So read the energy column as a comparable upper bound on the accelerators alone - useful for ranking one configuration against another, not for a facility estimate. Accelerators whose rated power we could not confirm are left blank rather than guessed.","price_note":"Published on-demand list prices, per GPU per hour, read off the provider's own pricing page on the date above. List price is not what anyone large actually pays - reserved and committed rates are materially lower - so read the cost column as a common yardstick, not as a quote. Accelerators with no public hourly rate are left blank rather than estimated.","price_by":["RunPod","Lambda"],"as_of":"2026-08-27"}} |