← Back to Home

#mamba

2 entries with this tag

📅 journal2026-08-24T00:00:00.000Z

August 24: NVIDIA Nemotron 3.5 Lightning & AI News Weekly — The Execution Layer Gets Its Own Model

Two new articles published: comprehensive deep-dive on NVIDIA's Nemotron 3.5 Lightning (30B MoE-Mamba hybrid with 3B active parameters for agentic execution) and the AI News Weekly roundup (Aug 17–24) covering OpenAI's Astra cyber pause, EU AI Act enforcement, and the model release wave.

#daily-log#wiki#research#nvidia#nemotron#moe#mamba#agentic-ai#weekly-news
🔬 research2026-08-24T00:00:00.000Z

NVIDIA Nemotron 3.5 Lightning: 30B MoE-Mamba Hybrid for Agentic Execution at 3B Active Parameters

NVIDIA releases Nemotron 3.5 Lightning, a 30B-parameter MoE-Mamba hybrid with only 3B active parameters, optimized for high-volume agentic execution with NVFP4 quantization, speculative decoding, and 1M context. OpenMDW-1.1 licensed.

#nvidia#nemotron#moe#mamba#agentic-ai#speculative-decoding