{"slug": "prime-intellect-releases-prime-mega-moe-trainable-fused-moe-kernel-3-11x-faster", "title": "Prime Intellect Releases Prime Mega MoE: Trainable Fused MoE Kernel, 3.11x Faster Forward Pass", "summary": "Prime Intellect released Prime Mega MoE, a backward pass for DeepSeek's fused Mega MoE kernel that makes the kernel trainable and delivers a 3.11x faster forward pass. The release speeds Qwen3 MoE training steps by up to 34% in Prime Intellect's prime-rl reinforcement learning framework.", "body_md": "Prime Intellect released Prime Mega MoE, a backward pass for DeepSeek’s fused Mega MoE kernel. It makes the kernel trainable and speeds Qwen3 MoE training steps by up to 34% in prime-rl.\n\nThe post [Prime Intellect Releases Prime Mega MoE: Trainable Fused MoE Kernel, 3.11x Faster Forward Pass](https://www.marktechpost.com/2026/10/29/prime-intellect-prime-mega-moe-fused-moe-kernel-training/) appeared first on [MarkTechPost](https://www.marktechpost.com).", "url": "https://wpnews.pro/news/prime-intellect-releases-prime-mega-moe-trainable-fused-moe-kernel-3-11x-faster", "canonical_source": "https://www.marktechpost.com/2026/10/29/prime-intellect-prime-mega-moe-fused-moe-kernel-training/", "published_at": "2026-10-29 18:28:57+00:00", "updated_at": "2026-10-29 19:49:29.348308+00:00", "lang": "en", "topics": ["machine-learning", "ai-research", "ai-infrastructure", "large-language-models", "developer-tools"], "entities": ["Prime Intellect", "Prime Mega MoE", "DeepSeek", "Qwen3", "prime-rl"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/prime-intellect-releases-prime-mega-moe-trainable-fused-moe-kernel-3-11x-faster", "markdown": "https://wpnews.pro/news/prime-intellect-releases-prime-mega-moe-trainable-fused-moe-kernel-3-11x-faster.md", "text": "https://wpnews.pro/news/prime-intellect-releases-prime-mega-moe-trainable-fused-moe-kernel-3-11x-faster.txt", "jsonld": "https://wpnews.pro/news/prime-intellect-releases-prime-mega-moe-trainable-fused-moe-kernel-3-11x-faster.jsonld"}}