NVIDIA NIM API: free access, limits and failover

What is free on NVIDIA

NVIDIA says members of the NVIDIA Developer Program have free access to NIM API endpoints for prototyping, and signing up enrols you. Usage runs on trial service credits under the NVIDIA API Trial Terms, and without a paid subscription it is limited to internal testing and evaluation, not production.

NVIDIA free-tier limits

NVIDIA does not publish credit amounts or rate-limit numbers. The trial terms say use may be limited by a number of API accesses or by duration, and that NVIDIA may offer ways to receive or buy more credits once they run out.

Get a NVIDIA API key

Open a model page, choose Get API Key and enter an email address, which enrols you in the NVIDIA Developer Program. Past a number of sessions, NVIDIA asks for personal information such as your name.

https://build.nvidia.com/

OpenAI compatibility

NVIDIA's model pages call the endpoints with the official OpenAI Python client, and the reference documents POST /v1/chat/completions. NVIDIA does not label the API "OpenAI-compatible" in so many words.

https://integrate.api.nvidia.com/v1

Data and privacy

The trial terms say NVIDIA does not store or use your content at the end of each session, and also that it collects content without identifying users to improve its products, including AI models, and logs use for security and abuse monitoring.

Through Ficelle, your NVIDIA key stays on your machine and prompts go straight to NVIDIA: there is no Ficelle proxy in between.

Before you rely on it

• Check the trial terms before attaching this endpoint to a deployed application: the unpaid allowance permits internal testing and evaluation only. Production requires a paid subscription.
• No official page publishes credit amounts or rate limits.
• Endpoints may be pre-release and withdrawn at any time.
• The trial terms allow inputs and outputs to be used to improve NVIDIA products.

Use NVIDIA with automatic failover

Ficelle can include NVIDIA behind your local OpenAI-compatible endpoint when the integration is enabled and its models pass free-access and capability checks. NVIDIA is listed in Ficelle Pro's maintained provider pack. Check your installed configuration, credentials and account entitlement before relying on it. Another eligible candidate must be available for failover to succeed. Local routing guards do not control upstream billing when an account permits paid usage beyond its free allocation. See strict-zero routing and account limits.

# NVIDIA ships in Ficelle Pro's maintained provider pack.
ficelle doctor --text          # check this install before configuring provider access

from openai import OpenAI

client = OpenAI(base_url="http://127.0.0.1:8646/v1", api_key="ficelle-local")

# Ficelle selects among configured, enabled, eligible models.
# If none can serve the request, it returns an error.
client.chat.completions.create(
    model="ficelle/auto-tools",
    messages=[{"role": "user", "content": "..."}],
)

See what Ficelle Pro includes →

Pick a model type

Choose a model type for your request. Eligibility depends on the installed provider configuration, catalog data, free-access evidence and capability checks. A catalog listing alone does not confirm that your account can call a model.

• Tool calling
• JSON & structured output
• Fast, low-latency
• Reasoning
• Coding assistants
• Agent orchestration
• Long context
• Context compression
• Vision
• Audio input
• Video input
• Multimodal

Questions