Nemotron 3 Ultra
NVIDIA · released Jun 3, 2026 · nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16
- Type
- Open weightsCustom licence
- Params
- 561B
- Context
- 512K
about 384K words of context · download allowed, licence restricts use
Our take
Written Aug 3, 2026Nemotron 3 Ultra is a large text-only model from NVIDIA with a 512,288-token request limit and a custom restricted licence. Its measured agent scores sit below zero, so it is a capacity-first choice rather than a quality-first one.
Pick this for very long documents or extended conversations that need half a million tokens in a single pass, or for high-throughput batch work where you can justify paying more for faster output. Use it if you are already inside the NVIDIA ecosystem and its licence terms are acceptable. Skip it if you need proven agentic performance, a permissive open licence, or any multimodal input.
The case for it
- 512,288-token request limit — among the largest we track for dense text work.
- Throughput varies sharply by host: the fastest measured offer runs almost three times quicker than the slowest.
- Lowest price in its own offer set is noticeably cheaper than its highest-price option.
The case against it
- Negative Arena Agent scores on both measured dates, firmly below the higher-is-better zero line.
- Active parameter count is undisclosed, so true per-token inference cost is unclear.
- Custom restricted licence, not Apache or MIT — commercial use is constrained by NVIDIA's terms.
How good is it?
IntelligencePuzzles, maths, exam questions
Nobody we watch has scored Nemotron 3 Ultra for this. We would take the rating from Arena Text (overall).
CodingWriting and fixing code on its own
Nobody we watch has scored Nemotron 3 Ultra for this. We would take the rating from Arena Coding.
AgenticPlanning, calling tools, staying on task
Arena Agent (IPS)34th of 36 · −0.131
Arena Agent (IPS) is the only board that has scored it for this.
WritingWe do not rate this
Nobody we watch has scored this model for writing. Two boards come close and neither tests writing: Arena Creative Writing asks people which of two replies they prefer, and LiveBench Language tests whether a model understood a passage.
Every published score for this model1 scoreEvery figure we hold, from 1 board, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
- Fits in memory
- weights load entirely on the card
- Spills to system RAM
- some weights offload; much slower
- Too large
- will not load even with offload
- est
- size is calculated; the verdict could change by 10%
GeForce RTX 4090 · 24 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M3 Ultra (80-core GPU) · 512 GB
Borderline fit on an estimated size. It leaves 17.9 GB spare on a size we calculated rather than measured, and a 10% error either way would change the answer.
Memory use by level
Against a 24 GB card.
All 3 sizes here are calculated, not measured. We hold no measured file for this model, so each size comes from the parameter count and every verdict above inherits that uncertainty.
Check against your own machine → · All 71 devices, with every size →
Or rent it from someone else
Cheapest of 6 live listings. Picked at the widest standard context we hold, within one quantisation slice, so the numbers beside it are a price one host actually charges.
- per 1M tokens
- $0.50 in / $2.20 out
- Context served
- 512K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenRouter | $0.50 / $2.20 | 512K | not measured | Unknown | Unknown | Unknown |
| DeepInfrafp4 | $0.50 / $2.20 | 262K | not measured | Unknown | Unknown | Unknown |
| DeepInfrafp4 | $0.50 / $2.20 | 262K | 31 tok/s | No | No | Confirmed |
| Basetenfp4 | $0.60 / $2.40 | 203K | 126 tok/s | No | No | Confirmed |
| Venice AIfp8 | $0.63 / $3.13 | 256K | 47 tok/s | No | No | Confirmed |
| Together AI | $0.60 / $3.60 | 512K | 87 tok/s | No | No | Confirmed |
DeepInfra
Together AIAcross the 6 listings we hold: 4 say they do not train on prompts, 0 say they do and 2 do not say. 4 appear in the zero-retention registry we check; the rest are unknown to us rather than confirmed either way.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
- ✓
- Supported
- ✗
- Not supported
- Not published
- host gave no parameter list
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenRouter | ✓ | ✓ | ✓ |
| DeepInfrafp4 | |||
| DeepInfrafp4 | ✓ | ✓ | ✓ |
| Basetenfp4 | ✓ | ✗ | ✗ |
| Venice AIfp8 | ✓ | ✓ | ✗ |
| Together AI | ✓ | ✓ | ✓ |
Tool calling: 5 of 6 listings say yes, 1 publishes no parameter list. JSON output: 4 of 6 listings say yes, 1 says no, 1 publishes no parameter list. Strict schema: 3 of 6 listings say yes, 2 say no, 1 publishes no parameter list.
When we formed this view
Dates behind this page
Prices last checked 3d ago
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- 1 of 6 listings publish no parameter list, so what their API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 2 of 6 listings do not say whether they train on prompts.
Licence and identifiers
What the licence allowsCustom licence, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Custom licence
This model ships custom license terms that don't map to a known template. We haven't parsed them, so commercial use, redistribution and derivatives are unverified — review the original terms before shipping.
Identifiers
- Hugging Face
- nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16
- Architecture
- Mixture of experts
- Modality record
- text->text
- Catalogue slug
- nvidia-nemotron-3-ultra