deepseek-v3.2 tokens per second

Live benchmark data for deepseek-v3.2 on ollama. Speed, latency, and reliability — benchmarked every ~10 minutes.

Latest TPS
30.1 tok/s
24h avg TPS
23.4 tok/s
TTFT
609ms
24h reliability
99%
Last tested
18s ago

Tokens per second over time

Time to first token over time

Speed is measured as output tokens ÷ generation time. Each benchmark fires one streaming chat completion with a fixed ~300-token prompt, every ~10 minutes. Full methodology →