Provider guide
Where to run gpt-oss-120b
23 live offers tracked — output prices vary 3.5× between the cheapest and most expensive host, so the provider choice matters as much as the model choice.
FASTEST MEASURED
919 tok/s
measured throughput · $0.35 input / $0.75 output per million tokens
OpenRouterCHEAPEST$0.037 in/1M$0.17 out/1M—131K—42m agoDeepInfra$0.037 in/1M$0.17 out/1M—131Kbfloat167h agoNovita AI$0.050 in/1M$0.25 out/1M—131K—7h agoSambaNova$0.22 in/1M$0.59 out/1M—131K—7h agoDeepInfrazero-retention$0.037 in/1M$0.17 out/1M40 tok/s131Kbf1640m agoCoreWeavezero-retention$0.030 in/1M$0.17 out/1M22 tok/s131Kfp440m agoNovita AIzero-retention$0.050 in/1M$0.25 out/1M102 tok/s131Kfp440m agoGoogle Vertex AIzero-retention$0.090 in/1M$0.36 out/1M210 tok/s131K—40m agoSiliconFlowzero-retention$0.050 in/1M$0.45 out/1M18 tok/s131Kfp840m agoDigitalOcean Gradientzero-retention$0.070 in/1M$0.49 out/1M34 tok/s128K—40m agoMancer 2zero-retention$0.10 in/1M$0.50 out/1M31 tok/s131Kfp87h agoBasetenzero-retention$0.10 in/1M$0.50 out/1M157 tok/s128Kfp440m agoPhalazero-retention$0.15 in/1M$0.60 out/1M80 tok/s131K—40m agoDeepInfraturbo tier$0.15 in/1M$0.60 out/1M129 tok/s131Kbf1640m agoTogether AIzero-retention$0.15 in/1M$0.60 out/1M148 tok/s131K—40m agoAmazon Bedrockzero-retention$0.15 in/1M$0.60 out/1M190 tok/s131K—40m agoAmazon Bedrockzero-retention$0.15 in/1M$0.60 out/1M122 tok/s131K—40m agoGroqzero-retention$0.15 in/1M$0.60 out/1M395 tok/s131K—40m agoNebius AI Studiozero-retention$0.15 in/1M$0.60 out/1M285 tok/s131Kfp47h agoCerebraszero-retention$0.35 in/1M$0.75 out/1M919 tok/s131Kfp1640m agoParasailzero-retention$0.10 in/1M$0.75 out/1M142 tok/s131Kfp440m agoMarazero-retention$0.15 in/1M$0.75 out/1M137 tok/s131K—40m agoSambaNovazero-retention$0.14 in/1M$0.95 out/1M325 tok/s131K—40m ago
DeepInfra
Novita AI
SambaNova
Google Vertex AI
DigitalOcean Gradient
Together AI
Amazon Bedrock
Groq
Nebius AI StudioFull specs, hardware verdicts and benchmarks on thegpt-oss-120b model page →