LaunchDevelopersOctober 29, 2026

Prime Intellect releases Prime Mega MoE training kernel

Read original source →marktechpost.com

Prime Intellect extended DeepSeek's fused Mega MoE kernel with an overlapping backward pass, making it usable for training rather than inference only. Full training steps on Qwen3 MoE models run 32-34% faster in prime-rl and 48-55% faster than TorchTitan.

1 source

More stories today

Open the live feed