EU people, which stack works best for you in terms...
# agents
e
EU people, which stack works best for you in terms or latency? Models and deployment
s
deepgram nova, gpt-4.1 mini, cartesia sonic-3
stay tuned for deepgram flux will shave off even more.
use deepgram EU endpoint via the plugin, use inference.TTS for cartesia sonic-3, same for openai (I tested both with Azure directly and via the inference it was the same speed). If you want to go even faster use gpt-oss-120b hosted on groq
e
thanks. have you tried gemini 2.0/2.5 flash?
s
its weak in prompt following in my opinion. Gpt 4.1-mini performs best