LaunchAI ModelsDevelopers
Jun 14, 12:26 PM
Xiaomi serves MiMo V2.5 at 1000-3000 tps with DFlash
Xiaomi's MiMo V2.5 achieves 1000-3000 tokens per second using DFlash and persistent kernel. The DFlash model is released, with open-source code promised soon.
·
Jun 14, 12:26 PM