Provider guide
Where to run Kimi K2.6
26 live offers tracked — output prices vary 1.6× between the cheapest and most expensive host, so the provider choice matters as much as the model choice.
FASTEST MEASURED
156 tok/s
measured throughput · $0.65 input / $3.41 output per million tokens
OpenRouterCHEAPEST$0.59 in/1M$2.48 out/1M—262K—14m agoBaidu$0.59 in/1M$2.48 out/1M35 tok/s262Kfp414m agoNovita AI$0.80 in/1M$3.40 out/1M—262K—6h agoModelRun$0.70 in/1M$3.40 out/1M111 tok/s262Kfp44d agoChutes$0.66 in/1M$3.50 out/1M5 tok/s262Kint414m agoDeepInfra$0.75 in/1M$3.50 out/1M—262Kfp46h agoStreamLake$0.85 in/1M$3.60 out/1M24 tok/s256Kfp814m agoAtlasCloud$0.95 in/1M$4.00 out/1M55 tok/s262Kint414m agoCloudflare Workers AI$0.95 in/1M$4.00 out/1M37 tok/s262K—14m agoSail Research$1.00 in/1M$4.00 out/1M—262Kfp87d agoDigitalOcean Gradientzero-retention$0.76 in/1M$3.20 out/1M56 tok/s262K—14m agoSiliconFlowzero-retention$0.77 in/1M$3.40 out/1M30 tok/s262Kfp814m agoDecartzero-retention$0.59 in/1M$3.40 out/1M34 tok/s262Kfp414m agoNovita AIzero-retention$0.80 in/1M$3.40 out/1M22 tok/s262K—14m agoInceptronzero-retention$0.60 in/1M$3.41 out/1M59 tok/s262Kint414m agoCoreWeavezero-retention$0.65 in/1M$3.41 out/1M156 tok/s262Kfp414m agoDeepInfrazero-retention$0.75 in/1M$3.50 out/1M19 tok/s262Kfp414m agoVenice AIzero-retention$0.75 in/1M$3.50 out/1M24 tok/s256Kint46h agoCrusoezero-retention$0.70 in/1M$3.50 out/1M35 tok/s262Kbf1614m agoParasailzero-retention$0.75 in/1M$3.50 out/1M17 tok/s262Kint414m agoBasetenzero-retention$0.95 in/1M$4.00 out/1M106 tok/s262Kfp414m agoMoonshot AIzero-retention$0.95 in/1M$4.00 out/1M20 tok/s262Kint414m agoFireworks AIzero-retention$0.95 in/1M$4.00 out/1M50 tok/s262K—20h agoSail Researchzero-retention$1.00 in/1M$4.00 out/1M31 tok/s262Kint414m agoTogether AIzero-retention$1.20 in/1M$4.50 out/1M18 tok/s262K—14m agoPhalazero-retention$1.09 in/1M$4.60 out/1M42 tok/s262K—14m ago
Full specs, hardware verdicts and benchmarks on theKimi K2.6 model page →