Week 23, 2026

A 30B total / 3B active parameter hybrid Mamba-2 + Transformer MoE language model built for efficient on-device and edge agentic tasks. It features a 1M-token context window, reasoning ON/OFF modes with configurable thinking budgets, and up to 4× faster throughput than its predecessor.

A 120B total / 12B active parameter hybrid Mamba-Transformer MoE language model with LatentMoE, multi-token prediction, and native NVFP4 pretraining. Optimized for complex multi-agent applications with a 1M-token context window and up to 5× higher throughput than the previous Nemotron Super.

A 4B-parameter multimodal, multilingual small language model designed as a robust content-safety moderator. It supports standard taxonomy safety classification and custom-policy enforcement with reasoning traces for text and image inputs.

NVIDIA's open foundation model for humanoid robot reasoning and control, combining an Eagle-based vision-language backbone with a diffusion transformer (DiT) action head for language-conditioned manipulation across diverse embodiments.

GLM-5-Turbo is a tuned variant of the GLM-5 series that prioritizes lower latency and efficient sequential execution. It shares the 200K context window and 128K output ceiling of the GLM-5 family and is positioned for agentic workflows that need many quick steps.

GLM-5V-Turbo is a vision-language variant of the GLM-5 family, built for multimodal coding, visual reasoning, and image-plus-text agent workflows. It offers a 200K context window and 128K maximum output.

MiniMax Speech 2.8 HD generates ultra-realistic, expressive speech with sound tags, supporting 40 languages, 7 emotions, and specified dialects for high-fidelity voice applications.

MiniMax Speech 2.8 Turbo balances speed and naturalness, supporting 40 languages, 7 emotions, and specified dialects for real-time, low-latency voice synthesis.

MiniMax Music 3.0 generates music from text and reference inputs with improved intent understanding, elevated sound quality, and more humanized vocals compared to the previous Music 2.0 generation.

An 8B-class physical-AI omni-model (8B reasoner + 8B generator) optimized for efficient inference on workstation-grade NVIDIA hardware such as the RTX PRO 6000. It unifies vision reasoning, world generation, and action prediction for robotics and physical AI prototyping.

A 4B-class physical-AI omni-model (2B reasoner + 2B generator) optimized for real-time robotic policy and visual reasoning at the edge. It unifies world generation, vision reasoning, and action prediction in a compact form factor for embedded deployment.