Frontier Retrieval Accuracy, Practical Deployment: NVIDIA Nemotron 3 Embed on FriendliAI

What is NVIDIA Nemotron 3 Embed?

NVIDIA Nemotron 3 Embed 8B and NVIDIA Nemotron 3 Embed 1B help teams build more accurate and efficient retrieval systems. As agents retrieve more often across unstructured documents, tools, code, knowledge bases, and memory systems, retrieval quality directly affects agent accuracy, latency, andtrust. If retrieval misses the right passage, code file, or policy, the downstream agent starts from the wrong context.

Nemotron 3 Embed addresses this with a practical accuracy-efficiency curve:

• Nemotron 3 Embed 8B — a frontier-quality embedding model designed to establish the accuracy ceiling across major retrieval benchmarks, topping the RTEB leaderboard.
• Nemotron 3 Embed 1B — an efficient embedding model designed to retain more than 95% of the 8B model's accuracy through pruning, distillation, and quantization-aware training — built for high-volume production workloads.

Both models handle text and code with a 16K token context length, and ship with open weights, datasets, recipes, and fine-tuning guidance, so teams can adapt them to domain-specific retrieval needs and deploy with full control over their data and infrastructure.

When to use Nemotron 3 Embed

Multi-turn agents retrieve repeatedly for planning, memory, code, and tool-use context. Weak retrieval increases turn count, token usage, and hallucination risk; strong retrieval keeps agents grounded. For retrieval to become a default agent capability, embedding models need to feel almost as easy to use as file search: fast, inexpensive, and accurate.

Nemotron 3 Embed is designed for exactly that:

• Agentic retrieval — a stronger retrieval layer for multi-turn planning, query decomposition, query rewriting, and repeated lookup loops
• RAG & enterprise search — grounding copilots and retrieval pipelines in enterprise documents and knowledge bases
• Code retrieval — retrieving relevant source files, functions, and implementation examples for developer copilots and software engineering agents

Running Nemotron 3 Embed on FriendliAI

FriendliAI is the Frontier Inference Cloud for Agents, delivering frontier-level intelligence, throughput, and lower cost of inference to complete agentic tasks.

Retrieval is where efficiency compounds. An agent may call an embedding model dozens of times per task, every query rewrite, memory lookup, and code search is another inference call. FriendliAI's inference stack is built to make exactly this kind of high-frequency, latency-sensitive workload fast and economical in production.

What FriendliAI brings to Nemotron 3 Embed:

High-throughput embedding inference FriendliAI's inference engine maximizes throughput per GPU through continuous batching and kernel-level optimizations, well suited to the bursty, high-volume query patterns of agentic retrieval, where a single agent turn can fan out into many embedding calls.

Cost-efficient retrieval at scale Retrieval only becomes a default agent capability when it's inexpensive enough to use repeatedly. FriendliAI delivers higher tokens-per-dollar than alternative OSS serving stacks, making semantic search practical to run as often as agents need it — almost like file search.

Production-grade reliability Retrieval sits on the critical path of every agent turn. FriendliAI provides enterprise-grade reliability with dedicated endpoints, auto-scaling, and operational monitoring — so retrieval never becomes the bottleneck in your agent loop.

Available on Friendli Dedicated Endpoints Both Nemotron 3 Embed 8B and 1B are available via Friendli Dedicated Endpoints: reserved GPU capacity with predictable latency for production retrieval workloads. Deploy in a few clicks and send embedding requests through an OpenAI-compatible API, on infrastructure that scales with your query volume.

Get Started

Nemotron 3 Embed is available on FriendliAI starting today via Friendli Dedicated Endpoints reserved GPU capacity for consistent, high-throughput production workloads where latency predictability matters.

To deploy NVIDIA Nemotron 3 Embed on Friendli Dedicated Endpoints:

1️⃣ Navigate to the dedicated endpoint creation page.

2️⃣ Choose your desired model: [nvidia/NVIDIA-Nemotron-3-Embed-8B-BF16] or [nvidia/NVIDIA-Nemotron-3-Embed-1B-BF16] or [nvidia/NVIDIA-Nemotron-3-Embed-1B-NVFP4]

3️⃣ Click "Create"

You can send embedding requests using any OpenAI-compatible API/SDK. Here’s how to set up:

Prerequisites

Before getting started, you'll need:

• A FriendliAI account
• A Friendli Token from Friendli Suite settings

Install the package

uv pip install friendli

Environment Setup

Set up your FriendliAI API key (aka Friendli Token):

export FRIENDLI_TOKEN="your-token-here"

Example Code

import os

from friendli import SyncFriendli

with SyncFriendli(
    token=os.environ["FRIENDLI_TOKEN"],
) as friendli:
    res = friendli.dedicated.embeddings.embeddings(
        model="your-dedicated-endpoint-id",
        input_="Hello, Nemotron 3 Embed!",
        encoding_format="float",
    )

    # Handle response
    print(res.data[0].embedding)

Deploy NVIDIA Nemotron at Scale with FriendliAI

FriendliAI is proud to partner with NVIDIA to offer the Nemotron family of models to the developer community and businesses building production-ready, agentic AI systems.

NVIDIA Nemotron 3 Embed brings frontier retrieval accuracy to practical enterprise deployment, better context, fewer wasted tokens, and more grounded AI systems. FriendliAI is your platform to take it to production.

👉 Launch your Nemotron 3 Embed Dedicated Endpoint today and give your agents a retrieval layer that keeps up with them.