avatar
Laughing🪁
14,089 Twitter followers
Follow
亚太主厨👨🏻‍🍳 @GoKiteAI | Prev: CMO @WeirdoGhostGang @outaxyz PDSA @AlibabaGroup Researcher @BlockBeatsAsia | 记录所思所感所想，观点仅代表个人
Posts
avatar
Laughing🪁
Thread
I wonder how many of you, like me, love using models from @GroqInc. The advantages are: 1️⃣ Extremely fast inference speed In my experience, as soon as the prompt is sent, the answers almost instantly flood the screen, reaching speeds of hundreds of tokens/second. Chatting and running agents feel like instant replies from a real person. Once you've used it, you can't go back. You can try it out in the playground: console.groq.com/playground My self-built AI voice input tool uses the Whisper model from Groq, and I feel the recognition speed is even faster than Typeless. The reason for the speed is that Groq uses an ASIC chip optimized for LLM inference tasks, achieving extremely low latency and extremely high deterministic speed. Those interested can look up the information themselves. 2️⃣ Accuracy is almost identical to the original model, with speed and stability far superior, making it incredibly comfortable to use. 3️⃣ The free quota is very attractive; you can get a bunch of models for free without binding a card, such as Llama and Qwen.
loading indicator
Loading..