Provider guide
Where to run DeepSeek V3.2
19 live offers tracked — output prices vary 14.5× between the cheapest and most expensive host, so the provider choice matters as much as the model choice.
FASTEST MEASURED
40 tok/s
measured throughput · $0.37 input / $1.11 output per million tokens
Baidu$0.21 in/1M$0.31 out/1M26 tok/s131Kfp852m agoStreamLake$0.21 in/1M$0.32 out/1M20 tok/s128Kfp852m agoAtlasCloud$0.26 in/1M$0.38 out/1M21 tok/s164Kfp852m agoDeepInfra$0.26 in/1M$0.38 out/1M—164Kfp47h agoNovita AICHEAPEST$0.27 in/1M$0.40 out/1M—164K—7h agoOpenRouter$0.27 in/1M$0.40 out/1M—164K—53m agoGMICloud$0.29 in/1M$0.43 out/1M28 tok/s164Kfp852m agoAlibaba Cloud$0.37 in/1M$1.11 out/1M40 tok/s131Kfp852m agoAlibaba Cloud$0.37 in/1M$1.11 out/1M34 tok/s131K—7d agoFriendli$0.50 in/1M$1.50 out/1M26 tok/s164K—52m agoSambaNova$3.00 in/1M$4.50 out/1M—33K—7h agoDeepInfrazero-retention$0.26 in/1M$0.38 out/1M12 tok/s164Kfp452m agoNovita AIzero-retention$0.27 in/1M$0.40 out/1M22 tok/s164Kfp852m agoSiliconFlowzero-retention$0.26 in/1M$0.42 out/1M25 tok/s164Kfp87h agoVenice AIzero-retention$0.33 in/1M$0.48 out/1M7 tok/s160K—52m agoPhalazero-retention$1.00 in/1M$1.00 out/1M7 tok/s164K—52m agoDigitalOcean Gradientzero-retention$0.42 in/1M$1.36 out/1M40 tok/s164K—52m agoGoogle Vertex AIzero-retention$0.56 in/1M$1.68 out/1M14 tok/s164K—52m agoSambaNovazero-retention$3.00 in/1M$4.50 out/1M38 tok/s33K—52m ago
Baidu
DeepInfra
OpenRouter
SambaNova
DigitalOcean Gradient
Google Vertex AIFull specs, hardware verdicts and benchmarks on theDeepSeek V3.2 model page →