I have noticed that since the announcement of 4o being free for everyone, that sometimes (and you never know when, or after how much you’ve spoken) mid-conversation my GPT will answer back with “server experiencing heavy load” message, where what I have spoken is then not recorded or saved. I then have to repeat myself and try to remember if it was a detailed response I was providing.
I have also noticed that while I wait for the new voice model to be released the connection time is much greater than before. With all due respect, something needs to change… as I am wondering if I’m the only one here that is failing to see how paying a monthly subscription fee makes us any different then any other ‘sheep in the herd.’
Has anyone else also experienced significant lag with 4o or these interrupted conversations due to heavy server load?
1 Like
I have noticed this as well.
On day 1 it was remarkably fast, now it’s on par ((still a little faster) with 4-turbo.
One way to reproduce this: Playground & compare the two.
I’m sure open AI (and azure) are trying their best.
We can watch the response time increase as we get close to the day peak around 8am US Pacific time.
For 3 trials of gpt-4o @ 2024-05-19 03:01AM:
Stat | Minimum | Maximum | Average |
---|---|---|---|
stream rate | Min: 69.8 | Max: 79.1 | Avg: 73.833 |
latency (s) | Min: 0.2859 | Max: 0.4549 | Avg: 0.364 |
total response (s) | Min: 3.5736 | Max: 4.1081 | Avg: 3.827 |
total rate | Min: 62.316 | Max: 71.636 | Avg: 67.114 |
response tokens | Min: 256 | Max: 256 | Avg: 256.000 |
For 3 trials of gpt-4o @ 2024-05-19 05:24AM:
Stat | Minimum | Maximum | Average |
---|---|---|---|
stream rate | Min: 40.8 | Max: 74.8 | Avg: 63.267 |
latency (s) | Min: 0.3563 | Max: 0.606 | Avg: 0.444 |
total response (s) | Min: 3.7767 | Max: 6.8546 | Avg: 4.807 |
total rate | Min: 37.347 | Max: 67.784 | Avg: 57.554 |
response tokens | Min: 256 | Max: 256 | Avg: 256.000 |
(an hour after model release, that total response was 2.7s)
Im getting this too but with gpt 4 as for some reason even tho im paying for it i still cant get GPT 4o.