Requesty

nemotron-3-ultra-nvfp4

Fireworks AI/🇺🇸 US/chat

Nemotron-3-Ultra-550B-A55B-NVFP4 is a frontier-scale large language model (LLM) trained by NVIDIA, designed to deliver strong agentic, reasoning, and conversational capabilities. It is optimized for the most demanding workloads, including complex multi-step agents, long-context analysis, and high-accuracy reasoning over code, math, and science. The model employs a hybrid Latent Mixture-of-Experts (LatentMoE) architecture, utilizing interleaved Mamba-2 and MoE layers, along with select Attention layers. Like the Super model, the Ultra model incorporates Multi-Token Prediction (MTP) layers for faster text generation and improved quality, and it is trained using an NVFP4 pre-training recipe to maximize compute efficiency. The model has 55B active parameters and 550B parameters in total.More
All models →

Which id to call

fireworks/nemotron-3-ultra-nvfp4

This exact deployment on Fireworks AI, with no routing and no failover. Send it as the model field.

Input /1M

$0.60

$0.12 cached

Output /1M

$2.40

4.0x input

Context

262K

262K output

Added

Jun 2026

chat

Capabilities 3/8

VisionReasoningTool callingCachingWeb searchJSON schemaComputer useImage generation

Provider rates

What nemotron-3-ultra-nvfp4 costs

Provider prices per 1M tokens, updated September 11, 2026.more

These are the upstream provider rates. Pay as you go adds 5%, or 0% if you bring your own keys, and there is no per-request fee. Prompt caching and routing change what you pay against these rates, not the rates themselves.

Input /1M

$0.60

Output /1M

$2.40

Cache write /1M

-

Cache read /1M

$0.12

What a workload costs

100K input + 10K output
$0.0840
1M input + 100K output
$0.84
10M input + 1M output
$8.40

At the rates above, before caching. A cache read costs $0.12 per 1M, so repeated context lands under these figures.

model=

Which id to call

One base url, https://router.requesty.ai/v1, and one key for every id here.

fireworks/nemotron-3-ultra-nvfp4

This exact deployment on Fireworks AI, with no routing and no failover. Send it as the model field.

OpenAI compatible

Call it in three lines

Change the base url, use your Requesty key, set the model to any id on the left. Existing OpenAI SDK code needs no other edit.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
from openai import OpenAI

client = OpenAI(
    api_key="YOUR_REQUESTY_API_KEY",
    base_url="https://router.requesty.ai/v1",
)

response = client.chat.completions.create(
    model="fireworks/nemotron-3-ultra-nvfp4",
    messages=[
        {"role": "user", "content": "Explain quantum computing in one paragraph."},
    ],
)

print(response.choices[0].message.content)

Reference

Specs and data terms

What the catalog reports for this deployment, and what the provider does with the traffic.

Context window262K tokens
Max output262K tokens
API typechat
AddedJun 2026
Model id
Data retentionNone
Used for trainingNo
Served from🇺🇸 US
Privacy policyFireworks Privacy Policy →

Public leaderboards

Benchmark scores

Benchmarks measure the model, not this endpoint, so they are the same wherever these weights are served.

No benchmarks are published for this exact variant yet.

Region-specific deployments and highspeed tiers usually share scores with their base model. Try the base model page or the Fireworks AI models overview.

Same provider

More from Fireworks AI

Newest first, on the same provider and the same key.

deepseek-v4.1-flash

1.0M ctx/$0.22/1M

deepseek-v4-flash-vision-exp

1.0M ctx/$0.22/1M

glm-5.3-flash

1.0M ctx/$0.15/1M

glm-5.3

1.0M ctx/$1.40/1M

nemotron-lightning-3.5-30b-a3b

262K ctx/$0.05/1M

deepseek-v4-pro-0813

1M ctx/$1.32/1M

Reference

nemotron-3-ultra-nvfp4 questions

How much does nemotron-3-ultra-nvfp4 cost?

nemotron-3-ultra-nvfp4 is priced at $0.60 per million input tokens and $2.40 per million output tokens when accessed via Requesty. Prompt caching is supported, which can cut effective input cost by up to 90% on repeated context. Those are the upstream provider rates: pay as you go adds 5% on top, or 0% if you bring your own provider keys.

What is the context window of nemotron-3-ultra-nvfp4?

nemotron-3-ultra-nvfp4 has a context window of 262K tokens, with a maximum output of 262K tokens per response. That's roughly 350 words of input you can fit in a single prompt.

What can nemotron-3-ultra-nvfp4 do?

nemotron-3-ultra-nvfp4 supports tool calling, extended reasoning, prompt caching. You can call it through any OpenAI-compatible client by pointing base_url to Requesty.

How do I use nemotron-3-ultra-nvfp4 with the OpenAI SDK?

Install the OpenAI SDK, set base_url to "https://router.requesty.ai/v1", set your API key to your Requesty key, and set the model to "fireworks/nemotron-3-ultra-nvfp4". The Quickstart above shows Python, JavaScript and cURL snippets.

Can I run nemotron-3-ultra-nvfp4 through Requesty?

Yes. nemotron-3-ultra-nvfp4 runs through Requesty's OpenAI-compatible API, served from Fireworks AI. You do not host the model yourself: point base_url at Requesty, set the model to "fireworks/nemotron-3-ultra-nvfp4", and requests are routed to the upstream provider with automatic failover. The same key gives you 600+ other models too.

Call nemotron-3-ultra-nvfp4 through one endpoint

One key for this endpoint and 600+ other models. No markup on provider prices, automatic failover, caching built in.

All Fireworks AI models