nvidia
NVIDIA Nemotron 3 Super API price cut 75% on input, 83% on output
NVIDIA's Nemotron 3 Super model saw its OpenRouter API prices drop sharply on November 7, with input costs falling 75% to $0.075 per million tokens and output costs falling 83.3% to $0.40 per million tokens compared to the previous day.
OpenRouter price, per 1M tokens
- $0.075
- $0.40
Price change details
The input price decreased from $0.30 to $0.075 per million tokens, a 75% reduction. The output price decreased from $2.40 to $0.40 per million tokens, an 83.3% reduction. The new prices are effective as of November 7, 2026, compared to November 6, 2026.
Model specifications
According to NVIDIA, the model has 120B total parameters and 12B active parameters using a LatentMoE architecture that combines Mamba-2, MoE, and Attention with Multi-Token Prediction. The context length supports up to 1M tokens. The minimum GPU requirement is 2× H100-80GB. Supported languages include English, French, German, Italian, Japanese, Spanish, and Chinese. The model was pre-trained on over 25T tokens with a pre-training data cutoff of June 2025 and post-training cutoff of February 2026. It is released under the NVIDIA Nemotron Open Model License.
Key takeaways
- Nemotron 3 Super API input price cut 75% to $0.075 per million tokens.
- Output price cut 83.3% to $0.40 per million tokens.
- Model features 120B total parameters with 12B active MoE architecture.