Provider guide
Where to run Llama 3.2 3B Instruct
4 live listings tracked — output prices vary 6.7× between the cheapest and most expensive host, so the provider choice matters as much as the model choice.
FASTEST MEASURED
88 tok/s
measured throughput · $0.051 input / $0.34 output per million tokens
Novita AI$0.030 in/1M$0.050 out/1M—33K—21 days agoOpenRouterCHEAPEST$0.050 in/1M$0.33 out/1M—131K—2 hours agoParasailzero-retention$0.050 in/1M$0.33 out/1M86 tok/s131Kbf162 hours agoCloudflare Workers AI$0.051 in/1M$0.34 out/1M88 tok/s80K—2 hours ago
Novita AIFull specs, hardware verdicts and benchmarks on the Llama 3.2 3B Instruct model page →