TTFT = how fast it starts · TPS / TPOT = how fast it types · ITL p95 + Jitter = how smooth the typing is · Latency = total wait · Tokens = answer length.
Lower is better for everything except TPS (higher is better).
MAC= MacBook Pro (M1)
PLUM= Raspberry Pi 4 8GB
BLUE= Raspberry Pi 4 8GB