Portkey
Log in

Welcome to Portkey Forum

Updated last year

GPT-4 is Getting Faster 🐇

VVrushank | Portkey
·
Notice GPT-4 getting faster? We did too. 🐇

Over the last 3 months, GPT-4 and GPT-3.5 latencies have more than halved - for both your regular requests and computationally complex requests with high token counts.

Check out the findings: https://blog.portkey.ai/blog/gpt-4-is-getting-faster/
L
V
3 comments
LLev
Thanks a lot for the great article and sharing your research
  • did you test openai or azure? if azure, which region?
  • did you track "time to first chunk"?
VVrushank | Portkey
  • Azure: This test is only for direct OpenAI API - but with your nudge, I think we should check Azure across different regions as well.
  • Time to first chunk: That's also great feedback. We didn't track it in this report, but should do for the next one.
VVrushank | Portkey
@Lev would you be interested in collaborating for the next report?
Add a reply
Sign up and join the conversation on Discord