---
id: "nvidia/nemotron-3-super-120b-a12b"
also: ["nvidia/nemotron-3-super-120b-a12b-20230311", "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8", "requesty/nvidia/nemotron-3-super-120b-a12b", "vercel/nvidia/nemotron-3-super-120b-a12b"]
cheapest_list_usd_per_m: "0.075 in, 0.4 out, at deepinfra/bf16 on openrouter"
cheapest_effective_usd_per_m: "0.0749 in, 0.3989 out, at deepinfra/bf16 on openrouter"
context: 1048576
max_output: 16384
input: ["text"]
output: ["text"]
released: "2026-03-11"
reasoning: ["medium", "low", "none"]
api: ["/api/v1/chat/completions", "/api/v1/completions", "/api/v1/messages", "/api/v1/responses"]
params: ["frequency_penalty", "include_reasoning", "logit_bias", "logprobs", "max_tokens", "min_p", "presence_penalty", "reasoning", "reasoning_effort", "repetition_penalty", "response_format", "seed", "stop", "structured_outputs", "temperature", "tool_choice", "tools", "top_k", "top_logprobs", "top_p"]
endpoints: 8
gateways: "openrouter (5), requesty (2), vercel (1)"
no_training: 7
trains: 1
retains_prompts: 1
coding: 37.7000
agentic: 1.7000
---
# NVIDIA: Nemotron 3 Super

NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. It delivers up to 7x higher throughput, providing fast, cost-efficient inference for agentic tasks. Additionally, a long context window gives the model long-term memory, preventing AI agents from losing focus on long, multi-step tasks and ensuring high-accuracy results. Fully open with weights, datasets, and recipes, Super allows easy customization and secure deployment anywhere.