Provider guide
Where to run GPT-3.5 Turbo (older v0613)
3 live listings tracked — output prices vary 2.0× between the cheapest and most expensive host, so the provider choice matters as much as the model choice.
FASTEST MEASURED
50 tok/s
measured throughput · $1.00 input / $2.00 output per million tokens
OpenRouter$1.00 in/1M$2.00 out/1M—4K—3 hours agoMicrosoft Azure AIzero-retention$1.00 in/1M$2.00 out/1M50 tok/s4K—3 hours agoOpenAICHEAPEST$3.00 in/1M$4.00 out/1M18 tok/s16K—3 hours ago
OpenRouterFull specs, hardware verdicts and benchmarks on the GPT-3.5 Turbo (older v0613) model page →