LLM Speed Test — AI Tokens Per Second

Measure real AI inference speed. How fast does the model stream tokens to you?

Quick prompts
Your key & prompt go straight to your endpoint — never to us.

📊 What Is TTFT?

TTFT (Time to First Token) measures how long after sending your prompt you receive the very first word back from the model. Lower TTFT = more responsive feel. Fast models typically deliver 300–800 ms TTFT depending on server load and network.

⚡ Tokens Per Second

Tokens per second (TPS) measures the output streaming speed of the model. Higher TPS = text appears faster on screen. GPT-4: ~30–60 TPS. Claude Sonnet: ~60–100 TPS. This is affected by server load, network latency, and output length.

🤖 LLM Speed Comparison

  • Claude Sonnet — ~60–100 TPS
  • Claude Haiku — ~120–180 TPS
  • GPT-4o — ~50–80 TPS
  • GPT-4o mini — ~100–140 TPS
  • Gemini Flash — ~80–120 TPS
  • Numbers vary by region and server load