Nemotron 3 Super (120B A12B) (Reasoning)

by NVIDIA

Nemotron 3 Super (120B A12B) (Reasoning) is a reasoning model from NVIDIA, released on March 11, 2026. It has a 262K-token context window and can generate up to 262K output tokens. API pricing is $0.09 per 1M input tokens and $0.4 per 1M output tokens. It ranks #94 of 220 models by Intelligence Score with a score of 26.7.

Nemotron 3 Super is a 120B total / 12B active parameter hybrid Mamba-Attention Mixture-of-Experts model optimized for agentic reasoning, coding, planning, tool calling, and long-context analysis. It introduces LatentMoE (projecting tokens into a compressed latent space for expert routing, enabling 4x more experts at the same inference cost), Multi-Token Prediction for native speculative decoding (up to 3x faster generation), and native NVFP4 pretraining on Blackwell. The hybrid architecture interleaves Mamba-2 layers for linear-time sequence processing with strategically placed Transformer attention layers as global anchors, supporting a 1M-token context window. Pre-trained on 25 trillion tokens and post-trained with multi-environment RL across 21 configurations using NeMo Gym/RL with 1.2 million rollouts. Achieves up to 5x higher throughput than previous Nemotron Super and 2.2x higher throughput than GPT-OSS-120B while maintaining comparable accuracy.

Key specs

Compare on the AI leaderboard · All AI models

More models from NVIDIA

• Nemotron 3 Ultra (550B A55B) (Reasoning) — NVIDIA
• Nemotron 3.5 Lightning (30B A3B) (Reasoning) — NVIDIA
• Nemotron 3 Nano (30B A3B) (Reasoning) — NVIDIA
• Llama 3.1 Nemotron Ultra 253B v1 — NVIDIA
• Nemotron Nano 9B v2 — NVIDIA
• Llama 3.1 Nemotron Nano 8B V1 — NVIDIA