all tiers / NVIDIA
build.nvidia.com API endpoints free tier limits (2026)
nvidia advertises 'free inference with leading models' on build.nvidia.com. the limits are not published on the pages we could read.
per minute
per day
per month
other limits / notes from sourceFree access to NIM API endpoints for prototyping/research/testing (not production)
card required
phone required
used for training
age req (as stated by the source)adult of legal age (api trial terms)
requirementsNVIDIA Developer Program
region restrictions
not verified: no official RPM/credit amounts; build.nvidia.com/faq 404
corrected 2026-10-18: api trial terms require an adult of legal age. still no official limits or credit amounts. source
models included
nvidia/nemotron-3.5-super-vl-120b-a12bnvidia/nemotron-3.5-lightning-30b-a3bdeepseek-ai/deepseek-v4.1-flashnvidia/nemotron-3-ultra-550b-a55b
notes
rate limits / credits not verified (js-rendered pages).
couldn't find these on the official page: card required, phone required, training on your data, region restrictions. if you can, send the link and we'll check it.