NVIDIA API
NVIDIA designs the GPUs and the AI computing platform behind much of the data-centre AI stack, and publishes open language models under the Nemotron family. On Felo, the NVIDIA entry currently exposes Nemotron 3 Ultra, free to call through the same Felo key as the rest of the catalog.
- Models
- 3
- chat ×3
- Input per 1M
- $0
- Output per 1M
- $0
- Largest context
- 1M
What NVIDIA provides
NVIDIA builds the GPU and AI-computing platform used across data centers, and its AI stack extends into software for building and running models. That includes the Nemotron family of open models, NeMo for building and optimizing agents, NIM inference microservices for deploying models, and a hosted API catalog for trying models before deploying them.
NVIDIA models on Felo
Felo currently lists one NVIDIA model, Nemotron 3 Ultra (free). It is an open reasoning model built for agentic work, coding, planning and long-context tasks, and the table on this page is generated from the live catalog so its status and pricing stay current.
Access and pricing
NVIDIA serves Nemotron through its NIM API catalog, including a free endpoint that is rate limited, while NIM microservices can be self-hosted for production workloads. On Felo, the NVIDIA entry is free to call: its published input and output rates are $0, so calls are not charged to your Felo balance. Check the model page for the current status.
Why use NVIDIA through Felo
The free Felo route lets you test Nemotron 3 Ultra without an NVIDIA API key or a separate NVIDIA account, using the same key and balance as the rest of the catalog. If you later need dedicated throughput or self-hosting, NVIDIA NIM provides that path on NVIDIA infrastructure.
Who NVIDIA is for
Nemotron 3 Ultra fits teams evaluating an open reasoning model for coding agents, deep research, multi-step planning and long-running enterprise workflows. It is a text-only route and Felo currently lists just this one NVIDIA model, so teams that need image, video or multimodal models should use other providers in the catalog; teams deploying at scale should look at NVIDIA NIM directly.
- Input per 1M
- $0
- Output per 1M
- $0
- Context
- 1M
- Max output
- 131.072K
- Input per 1M
- $0
- Output per 1M
- $0
- Context
- 1M
- Max output
- 65.536K
- Input per 1M
- $0
- Output per 1M
- $0
- Context
- 1M
- Max output
- 65.536K
| Model | Input per 1M | Output per 1M | Context | Max output |
|---|---|---|---|---|
| $0 | $0 | 1M | 131.072K | |
| $0 | $0 | 1M | 65.536K | |
| $0 | $0 | 1M | 65.536K |
Prices are published standard rates in USD per 1M tokens, before channel discounts.
Frequently asked questions
- What is NVIDIA?
- NVIDIA designs GPUs and the AI computing platform used in data centers. Its AI stack includes the open Nemotron models, NeMo for agents, NIM inference microservices and a hosted API catalog for testing and deploying models.
- Which NVIDIA models can I call through Felo?
- Felo currently lists one NVIDIA model, Nemotron 3 Ultra (free). The live model table on this page is authoritative.
- Do I need an NVIDIA account or API key?
- No. The NVIDIA model on this page is called with a Felo API key, so no separate NVIDIA account or NVIDIA API key is required.
- Is NVIDIA free on Felo?
- Yes. Nemotron 3 Ultra is published on Felo at $0 for input and output tokens, so calls are not charged to your balance. NVIDIA's own free endpoint is rate limited, and the Felo table is the authoritative current rate.
- What is Nemotron 3 Ultra best for?
- Coding agents, deep research, multi-step planning and long-running enterprise agent workflows. It is a text-only model, so image or multimodal work needs a different provider.