3 ms·

Ask HN: Slow OpenAI Inference on AWS Bedrock

2 points by timedude 17d ago

Inference on AWS Bedrock for OpenAI models has been very slow and glitchy the last few days with today being the worst. Very long latency of over 5 minutes to first token. What's going on?

timedude 17d ago
us-east-1 seems to be the worst region right now.
mikert89 16d ago
likely they sold all the gpu capacity to enterprise clients