Llama3V is suspected to have been stolen from the MiniCPM-Llama3-v2.5 projectgithub.com·30 pts·zinccat·7
DeepSeek V2, near GPT4 performance with 30% GPT3.5 cost ($0.14M input tokens)deepseek.com·16 pts·zinccat·2
SEQUOIA: Exact Llama2-70B on an RTX4090 with half-second per-token latencyinfini-ai-lab.github.io·131 pts·zinccat·61