side by side Groq vs. GPT-3.5, completely different user experience, a game changer for products that require low latency

Can be a game changer in the voice interaction products. Last time I checked vocode, vapi and other voice interaction production were adding filler words to deal with high latency.

Feels like when i switched from EDGE to 3G on my phone 20 years ago

LPUs for the win 🥇

low latency is a universal need!

I tried it out. Sure the answer is very fast, but you need to mention, that you need to wait until your request is started, because you land in a queue. I waited in my first try 14 Seconds for the answer.

How's the quality diff?

The side-by-side comparison is visually misleading. The two runs are technically different in terms of model sizes, connections, response display. Groq should be compared against an NVIDIA GPU chip, not to compare with an application like GPT-3.5

@HPC_Guru Laser vs Dot matrix printer

we're at the dawn of Language Processing Units applications

Now do a cost per token comparison