Hi everyone I built this tool after noticing something interesting: AI response speed isn’t consistent across regions. The same model, same prompt — but very different Time To First Token depending on where the request originates. This project started as a research experiment to measure real-world LLM latency under different network conditions. Instead of relying on lab environments, I wanted to capture actual user-side performance across countries and ISPs. What this tool does: • Measures TTFT (Time To First Token) • Measures total response latency • Evaluates stability across repeated calls • Takes about 3 minutes to run What it does NOT do: • No prompt storage • No personal data collection • No background tracking If you decide to run it, I’d love to hear: Your country Whether the results surprised you If you noticed instability during longer responses Curious to see how different regions compare 🌍 Happy to answer any technical questions about the measurement method.

LLM Region Speed Tester

Measure real-world AI response speed from your country in 3m

Measure real-world AI response speed from your country in 3m