Calculate the TPS of the Jetson Orin NX 16 GB on local AI models

NVIDIA 16 GB LPDDR5 102 GB/s February 2023

Every model in our catalogue assessed against this card at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from this card's memory bandwidth and the size of each model once compressed.

Calculated for this card

432 models it can run

679 models in our catalogue altogether

Largest model it holds

Nemotron 3-Nano-30B-A3B

31.6B · Q3_K_M · 20.6 tok/s

Fastest model

Gemma 3 QAT 1B

43.4 tok/s · 1B

Which AI models can run on a Jetson Orin NX 16 GB?

Set the inputs, read the answer

More context means more memory for the conversation cache. Speed is for a fresh conversation and does not change with this setting.

Hides models that would only fit by being compressed below this point.

432 models match

Calculating
Quantisation Fit
43.4 tok/s

37–52

Gemma 3 1B 1B Mar 2025 1.8 GB 33k tokens Q8_0 Comfortable
43.4 tok/s

37–52

Gemma 3 QAT 1B 1B Apr 2025 1.8 GB 33k tokens Q8_0 Comfortable
43.4 tok/s

26–69 · low confidence

HGRN 1B (WT 103) 1B Nov 2023 1.8 GB 131k tokens ? Q8_0 Comfortable
43.4 tok/s

26–69 · low confidence

LLama 3..2 Typhoon 2 1B 1B Dec 2024 1.8 GB 131k tokens ? Q8_0 Comfortable
43.4 tok/s

26–69 · low confidence

OLMo-1B 1B Feb 2024 1.8 GB 131k tokens ? Q8_0 Comfortable
43.4 tok/s

26–69 · low confidence

Pythia-1b 1B Apr 2023 1.8 GB 131k tokens ? Q8_0 Comfortable
40.2 tok/s

24–64 · low confidence

OpenELM-1.1B 1.1B May 2024 1.9 GB 131k tokens ? Q8_0 Comfortable
39.4 tok/s

24–63 · low confidence

DeciCoder-1B 1.1B Aug 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
39.4 tok/s

24–63 · low confidence

SantaCoder 1.1B Jan 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
39.4 tok/s

24–63 · low confidence

TinyLlama-1.1B (1T token checkpoint) 1.1B Oct 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
39.4 tok/s

24–63 · low confidence

TinyLlama-1.1B (3T token checkpoint) 1.1B Oct 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
36.1 tok/s

22–58 · low confidence

EXAONE 4.0 (1.2B) 1.2B Jul 2025 2.0 GB 131k tokens ? Q8_0 Comfortable
36.1 tok/s

22–58 · low confidence

MinerU2.5 1.2B Sep 2025 2.0 GB 131k tokens ? Q8_0 Comfortable
36.1 tok/s

22–58 · low confidence

Pleias 1.0 1.2B 1.2B Dec 2024 2.0 GB 131k tokens ? Q8_0 Comfortable
36.1 tok/s

22–58 · low confidence

Pleias-RAG-1B 1.2B Apr 2025 2.0 GB 131k tokens ? Q8_0 Comfortable
35.3 tok/s

30–42

Llama 3.2 1B 1.2B Sep 2024 2.2 GB 131k tokens Q8_0 Comfortable
34.8 tok/s

21–56 · low confidence

MiniCPM-1.2B 1.2B Jun 2024 2.0 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

DeepSeek Coder 1.3B 1.3B Jan 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

DeepSeek-VL-1.3B 1.3B Mar 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

DigiRL 1.3B Jun 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

GLA Transformer 1.3B 1.3B Aug 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

Janus 1.3B 1.3B Oct 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

Kosmos-2.5 1.3B Aug 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

Otter 1.3B May 2023 2.1 GB 131k tokens ? Q8_0 Comfortable
33.4 tok/s

20–53 · low confidence

Phi-1 1.3B Oct 2023 2.1 GB 131k tokens ? Q8_0 Comfortable

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Jetson Orin NX 16 GB full specification

Everything on record for this board, ordered by how much it bears on running a language model rather than by how a spec sheet would list it. Memory comes first because it decides the outcome; the rest is context.

Memory

The two specifications that decide what this card can run and how quickly. Capacity sets which models fit; bandwidth sets how many tokens per second they produce once they do.

Memory size
16 GB
Memory bandwidth
102 GB/s
Memory type
LPDDR5
Memory bus width
128 bit
Memory clock
800 MHz

The chip

Which processor is on the board and how it was manufactured. A smaller process size generally means more performance for the same power.

Graphics processor
GA10B
Architecture
Ampere
Generation
Tegra(Ampere)
Foundry
Samsung
Process size
8 nm
Die size
200 mm²
Released
1 February 2023

Clock speeds

How fast the processor runs. Worth far less here than on a gaming benchmark: generating text is limited by memory bandwidth, so a higher clock barely moves the result.

Base clock
918 MHz
Boost clock
918 MHz

Processing units

What the chip contains. These drive graphics performance and matter mainly for processing a long prompt rather than for producing the answer.

Shading units
1,024
Texture mapping units
32
Render output units
16
Streaming multiprocessors
8
Tensor cores
32
L1 cache
128 KB
L2 cache
4 MB

Theoretical performance

Peak arithmetic rates published for the board. These are ceilings that no real workload reaches, and generating text reaches a small fraction of them because it is limited by memory rather than arithmetic.

Half precision (FP16)
3.8 TFLOPS
Single precision (FP32)
1.9 TFLOPS
Pixel rate
15 GPixel/s
Texture rate
29 GTexel/s

The board

What it takes to physically install and power the card — the practical constraints that decide whether it fits the machine you already own.

Power draw (TDP)
25 W
Bus interface
PCIe 4.0 x4
Slot width
IGP
Dimensions
70 mm

Software support

Which graphics and compute interfaces the card supports. CUDA compute capability is the one that bears on inference: below 7.0 there are no tensor cores, and modern inference software falls back to slower code paths.

CUDA compute capability
8.7
DirectX
12.2
OpenGL
4.6
Vulkan
1.4
OpenCL
3.0
Shader model
6.8

Listings

Where to buy a Jetson Orin NX 16 GB

No vendor is currently listing this card. Listings come from vendors who publish them here directly — browse the vendor directory to see who is selling what.

What the numbers mean

What the memory subsystem means for AI

Memory

16 GB

Bandwidth

102 GB/s

Largest model

Nemotron 3-Nano-30B-A3B

16 GB of LPDDR5 puts the Jetson Orin NX 16 GB comfortably into small and mid-sized models, with roughly 14.4 GB usable once the driver overhead is taken out. The largest models are out of reach without splitting them.

At 102 GB/s across a 128-bit bus, bandwidth is this card's real constraint. Every token requires reading the entire model out of memory, so a large model will feel slow here even when it fits.

The figure is the memory clock — 800 MHz here — multiplied by the bus width. It is why core counts predict generation speed so poorly.

The practical ceiling is Nemotron 3-Nano-30B-A3B at 31.6B, held at Q3_K_M and running at roughly 20.6 tokens per second.

The chip and how it was built

The Jetson Orin NX 16 GB is built on the GA10B graphics processor, using NVIDIA's Ampere architecture, as part of the Tegra(Ampere) generation.

The chip is manufactured by Samsung, on a 8 nm process, with a die measuring 200 mm². A smaller process generally means more performance for the same power, though for language models it matters far less than the memory subsystem.

It was released in February 2023, roughly 3 years ago. Inference software support tends to follow hardware by a year or two, so a card of this age generally has mature, well-optimised code paths available to it.

Compute throughput, and why it matters less than it looks

FP16

3.8 TFLOPS

Tensor cores

32

On paper the Jetson Orin NX 16 GB reaches 3.8 TFLOPS at half precision and 1.9 TFLOPS at single precision. These are peak figures no real workload sustains, and generating text reaches only a small fraction of them — decoding is limited by memory rather than arithmetic, which is why a card can look enormously powerful here and still produce tokens at an ordinary rate.

The card carries 32 tensor cores across 8 streaming multiprocessors. These accelerate the matrix arithmetic at the heart of a transformer, and they are what make prompt processing — reading a long document before answering — dramatically faster than it would otherwise be.

Clocks run from 918 MHz at base to 918 MHz boosted. Worth far less here than on a gaming benchmark: raising the clock speeds up the arithmetic, and the arithmetic is not what generation is waiting on.

Cache and processing units

The Jetson Orin NX 16 GB has 128 KB of L1 cache, backed by 4 MB of L2. Cache absorbs a share of the memory traffic that would otherwise hit the main bus, which is the one place on this page where a number other than bandwidth quietly affects generation speed — a large L2 lets more of the working set stay close to the cores.

There are 1,024 shading units, 32 texture mapping units, and 16 render output units. These drive graphics workloads and contribute to prompt processing, but they sit idle for much of the time a model spends generating a reply.

Power, size and installation

Power draw

25 W

The Jetson Orin NX 16 GB is rated at 25 W. Running a language model keeps a card busy in bursts rather than continuously — it draws hard while generating and idles between requests — so sustained draw over a working day is usually well below the rated figure.

The board occupies a igp, measuring 70 mm long. Worth checking against the case and power supply already in the machine, since the largest cards need considerably more of both than a typical desktop provides.

It connects over PCIe 4.0 x4. The interface governs how quickly a model is loaded from disk into the card, not how fast it runs once there, so a narrower link costs a few seconds at startup and nothing thereafter.

The extremes

The largest AI models that run on a Jetson Orin NX 16 GB

The biggest open-weight models that fit on this card, newest first. Each is shown at the best compression the card can hold.

  1. 01 North Mini Code 30B · Q3_K_M · Jun 2026 21.7 tok/s
  2. 02 Qwen 3.6-27B 27B · Q3_K_M · Apr 2026 4.3 tok/s
  3. 03 Qwen3.5-27B 27B · Q3_K_M · Feb 2026 4.3 tok/s
  4. 04 Nemotron 3-Nano-30B-A3B 31.6B · Q3_K_M · Dec 2025 20.6 tok/s
  5. 05 Nomos 1 30B · Q3_K_M · Dec 2025 21.7 tok/s
  6. 06 C2S-Scale 27B · Q3_K_M · Oct 2025 4.3 tok/s
  7. 07 Gemma-SEA-LION-v4-27B-IT 27B · Q3_K_M · Aug 2025 4.3 tok/s
  8. 08 ERNIE-4.5-VL-28B-A3B 28B · Q3_K_M · Jun 2025 23.2 tok/s
  9. 09 Qwen3-30B-A3B 30B · Q3_K_M · Apr 2025 21.7 tok/s
  10. 10 Gemma 3 QAT 27B 27B · Q3_K_M · Apr 2025 4.3 tok/s

The fastest AI models on a Jetson Orin NX 16 GB

Where this card produces tokens quickest. Smaller models dominate here, because generating each token means reading the whole model out of memory once.

  1. 01 Gemma 3 QAT 1B 1B · Q8_0 · 1.8 GB 43.4 tok/s
  2. 02 Gemma 3 1B 1B · Q8_0 · 1.8 GB 43.4 tok/s
  3. 03 LLama 3..2 Typhoon 2 1B 1B · Q8_0 · 1.8 GB 43.4 tok/s
  4. 04 OLMo-1B 1B · Q8_0 · 1.8 GB 43.4 tok/s
  5. 05 HGRN 1B (WT 103) 1B · Q8_0 · 1.8 GB 43.4 tok/s
  6. 06 Pythia-1b 1B · Q8_0 · 1.8 GB 43.4 tok/s
  7. 07 OpenELM-1.1B 1.1B · Q8_0 · 1.9 GB 40.2 tok/s
  8. 08 TinyLlama-1.1B (1T token checkpoint) 1.1B · Q8_0 · 1.9 GB 39.4 tok/s
  9. 09 TinyLlama-1.1B (3T token checkpoint) 1.1B · Q8_0 · 1.9 GB 39.4 tok/s
  10. 10 DeciCoder-1B 1.1B · Q8_0 · 1.9 GB 39.4 tok/s

Step by step

How to work out the tokens per second of a Jetson Orin NX 16 GB

You do not have to calculate anything by hand — the gputps.com calculator on this page has already worked it out for every model this card can hold. Reading off the answer takes six steps.

  1. 01

    Find the model in the table

    All 432 models the Jetson Orin NX 16 GB handles are already listed. The search box takes a name or a size such as 27b, which matches on parameter count.

  2. 02

    Set the context length you will actually use

    Drag the slider to the conversation length you plan to work at. The cache grows with the conversation, and on 16 GB it is often what pushes a large model over the edge.

  3. 03

    Set a minimum quality if you need one

    Each model is shown at the best compression this card can hold. A minimum quality hides the ones that only fit by being squeezed further than you would accept.

  4. 04

    Read the speed and the range

    Each speed is an estimate for a single conversation, with a range beneath it — 43.4 tok/s on Gemma 3 QAT 1B at the top end here. The same card and model vary by thirty to fifty per cent between inference runtimes.

  5. 05

    Read the fit verdict last

    A tight fit runs but leaves no room to raise the context later; comfortable has headroom. The memory column shows what each model needs against the 16 GB available.

  6. 06

    Open the model to compare cards

    Following a model through to its own page lists all the hardware that can run it, so you can see where the Jetson Orin NX 16 GB sits against the alternatives.

Answers

Jetson Orin NX 16 GB — common questions

01

Can a Jetson Orin NX 16 GB run a 30B model?

Yes. For example a Jetson Orin NX 16 GB runs ERNIE-4.5-VL-28B-A3B at Q3_K_M, using about 12.9 GB of memory and generating around 23.2 tokens per second.

02

How much memory does a Jetson Orin NX 16 GB have?

A Jetson Orin NX 16 GB has 16 GB of LPDDR5 memory. Around a tenth of that is reserved by the inference runtime and the driver, leaving roughly 14.4 GB available for a model and its conversation.

03

What is the memory bandwidth of a Jetson Orin NX 16 GB?

The Jetson Orin NX 16 GB has 102 GB/s of memory bandwidth, across a 128-bit memory bus. This is the single best predictor of how fast it generates text, because producing each token means reading the entire model out of memory once.

04

What type of memory does a Jetson Orin NX 16 GB use?

It uses LPDDR5 clocked at 800 MHz. HBM types are found on datacentre accelerators and carry far more bandwidth than the GDDR used on desktop cards, which is why they generate tokens considerably faster at the same capacity.

05

Who makes the Jetson Orin NX 16 GB?

The Jetson Orin NX 16 GB is a NVIDIA product, with the chip manufactured by Samsung, on a 8 nm process.

06

When was the Jetson Orin NX 16 GB released?

The Jetson Orin NX 16 GB was released in February 2023.

07

How much power does a Jetson Orin NX 16 GB use?

The Jetson Orin NX 16 GB has a rated board power of 25 W. Generating text draws hard in bursts and idles between requests, so average consumption over a working session is normally well below the rated figure.

08

How much cache does a Jetson Orin NX 16 GB have?

The Jetson Orin NX 16 GB has 128 KB of L1 cache, and 4 MB of L2 cache. Cache absorbs part of the memory traffic that would otherwise reach the main bus, so a larger L2 gives a modest lift to generation speed beyond what bandwidth alone predicts.

09

What are the TFLOPS of a Jetson Orin NX 16 GB?

The Jetson Orin NX 16 GB is rated at 3.8 TFLOPS at half precision and 1.9 TFLOPS at single precision. These are peak arithmetic ceilings rather than achievable rates, and text generation reaches only a small fraction of them because it is limited by memory bandwidth instead.

10

How many tensor cores does a Jetson Orin NX 16 GB have?

The Jetson Orin NX 16 GB has 32 tensor cores across 8 streaming multiprocessors. They accelerate the matrix arithmetic a transformer is built from, which mainly speeds up processing a long prompt rather than producing the reply.

11

Does the Jetson Orin NX 16 GB support CUDA?

Yes. The Jetson Orin NX 16 GB reports CUDA compute capability 8.7. Capability 7.0 and above has tensor cores, which modern inference software uses; below that it falls back to slower code paths for quantised models.

12

What bus interface does the Jetson Orin NX 16 GB use?

It uses PCIe 4.0 x4. This governs how fast a model is loaded onto the card rather than how fast it runs once loaded, so it costs a few seconds at startup and nothing during generation.

13

Is the Jetson Orin NX 16 GB good for running local AI models?

Its memory covers small and mid-sized models, though the largest are out of reach though its bandwidth means generation will feel slow on larger models. In total it runs 432 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.

14

Can a Jetson Orin NX 16 GB run a model that does not fit in its memory?

Only partly. Layers beyond the 16 GB sit in system memory and run at a fraction of the speed, so a mostly-offloaded model is rarely worth using. Every figure here assumes it is fully resident on the card.

15

Would two Jetson Orin NX 16 GB cards be twice as fast?

Capacity adds, throughput does not. Two of them give you 32 GB to work with rather than twice the tokens per second — every figure here is for a single Jetson Orin NX 16 GB.

16

What AI models can a Jetson Orin NX 16 GB run?

432 of the 679 open-weight language models we track fit on a Jetson Orin NX 16 GB and can be run locally on it. The table on this page lists every one, with the memory it needs, the quantisation it runs at and an estimated generation speed.

17

What is the largest AI model a Jetson Orin NX 16 GB can run?

The largest model in our catalogue that fits on a Jetson Orin NX 16 GB is Nemotron 3-Nano-30B-A3B at 31.6B parameters, compressed to Q3_K_M. It generates roughly 20.6 tokens per second and needs about 14.4 GB of the card's memory.

18

How many tokens per second does a Jetson Orin NX 16 GB produce?

It depends on the model. On a Jetson Orin NX 16 GB the fastest model we track is Gemma 3 QAT 1B at about 43.4 tokens per second, while larger models run proportionally slower because each token requires reading the whole model out of memory once. Speeds are estimates for a single conversation at a time.

19

Can a Jetson Orin NX 16 GB run a 7B model?

Yes. For example a Jetson Orin NX 16 GB runs Multi-Token Prediction 7B at Q8_0, using about 7.9 GB of memory and generating around 6.5 tokens per second.

20

Can a Jetson Orin NX 16 GB run a 13B model?

Yes. For example a Jetson Orin NX 16 GB runs DeepSeekMoE-16B at Q6_K, using about 13.7 GB of memory and generating around 21.9 tokens per second.

The other direction

Looking at it from the other side?

This page starts from the hardware. If you already know which model you want and need to know what it takes to run it, start from the model instead.

All GPUs