NVIDIA models
Every model we track from NVIDIA, newest release first, with its status, any shutdown date and its standard price per 1M tokens. The latest release months come first; older ones load below. Open a model for its full price table, price history and changelog.
Compare what each model supports on Features; see where each alias points on Aliases and which models are free to call on Free.
All models
Newest release first| Provider | Model | Released | Status | Input | Output | Cache hit |
|---|---|---|---|---|---|---|
| March 2026 | ||||||
| NVIDIA | nemotron-3-super-120b-a12b | Wed, Mar 11 | Active | — | — | — |
Show models with no release date (58)
USD per 1M tokens, standard tier. Cache hit is the price for input read from the provider's prompt cache. Where a model has prompt-size bands, the lowest band is shown; every band is on the model's page. Released is the day the model came out, or just the month when that's all that's known.