According to a recent LinkedIn post from FriendliAI, the company is featuring NVIDIA’s Nemotron 3 Super model on its platform from day one, with Dedicated Endpoints aimed at multi‑agent and specialized agentic AI applications. The post highlights a hybrid Mamba–Transformer architecture designed to improve token generation efficiency, reasoning speed, and response quality.
Claim 55% Off TipRanks
- Unlock hedge fund-level data and powerful investing tools for smarter, sharper decisions
- Discover top-performing stock ideas and upgrade to a portfolio of market leaders with Smart Investor Picks
The company’s LinkedIn post notes that Nemotron 3 Super targets leading benchmark performance across GPQA Diamond, AIME 2025, LiveCodeBench, IFBench, and BFCL, and introduces a “thinking budget” intended to control inference costs and avoid unnecessary compute. It also references large scale characteristics, including 120B total parameters with 12B active and context lengths up to 1M tokens.
As shared in the post, FriendliAI positions its infrastructure with a 99.99% uptime service level for running mission‑critical workloads using models like Nemotron 3 Super. For investors, this emphasis on reliability and high‑end model support may signal a focus on enterprise‑grade, performance‑sensitive deployments that can command premium pricing.
The integration of a flagship NVIDIA model could strengthen FriendliAI’s competitive stance within the AI infrastructure and inference platform segment, particularly for customers building complex agentic systems. If the company can translate early access to advanced models and benchmark performance into recurring enterprise usage, it may enhance revenue visibility and improve its positioning relative to other inference and serving providers.