Aria TPS calculator

Open weights Rhymes AI 24.9B parameters October 2024

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

Calculated for this model

241 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Xeon Phi 7120P

16 GB · IQ4_XS · 9.6 tok/s

Fastest card

B200

136 tok/s · 180 GB

Which GPUs can run Aria?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

241 cards match

Calculating

					Needs	Quantisation	Fit
136 tok/s 82–218 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	27.4 GB	Q8_0	Comfortable
136 tok/s 82–218 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	27.4 GB	Q8_0	Comfortable
109 tok/s 65–174 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	27.4 GB	Q8_0	Comfortable
109 tok/s 65–174 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	27.4 GB	Q8_0	Comfortable
86.9 tok/s 52–139 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	27.4 GB	Q8_0	Comfortable
83.2 tok/s 50–133 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	27.4 GB	Q8_0	Comfortable
83.2 tok/s 50–133 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	27.4 GB	Q8_0	Comfortable
79.6 tok/s 48–127 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	27.4 GB	Q8_0	Comfortable
70.7 tok/s 42–113 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	27.4 GB	Q8_0	Comfortable
70.7 tok/s 42–113 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	27.4 GB	Q8_0	Comfortable
70.7 tok/s 42–113 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	27.4 GB	Q8_0	Comfortable
67.0 tok/s 40–107 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	27.4 GB	Q8_0	Comfortable
57.2 tok/s 34–91 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	27.4 GB	Q8_0	Comfortable
57.2 tok/s 34–91 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	27.4 GB	Q8_0	Comfortable
57.2 tok/s 34–91 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	27.4 GB	Q8_0	Comfortable
57.2 tok/s 34–91 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	27.4 GB	Q8_0	Comfortable
57.2 tok/s 34–91 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	27.4 GB	Q8_0	Comfortable
47.2 tok/s 28–76 · low confidence	Tesla V100 SXM2 16 GB NVIDIA	16 GB	1,130 GB/s	Nov 2019	14.3 GB	IQ4_XS	Tight
43.5 tok/s 26–70 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	27.4 GB	Q8_0	Comfortable
43.5 tok/s 26–70 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	27.4 GB	Q8_0	Comfortable
40.1 tok/s 24–64 · low confidence	GeForce RTX 5080 NVIDIA	16 GB	960 GB/s	Jan 2025	14.3 GB	IQ4_XS	Tight
37.5 tok/s 22–60 · low confidence	Tesla V100 DGXS 16 GB NVIDIA	16 GB	897 GB/s	Mar 2018	14.3 GB	IQ4_XS	Tight
37.5 tok/s 22–60 · low confidence	Tesla V100 PCIe 16 GB NVIDIA	16 GB	897 GB/s	Jun 2017	14.3 GB	IQ4_XS	Tight
37.5 tok/s 22–60 · low confidence	Tesla V100 SXM2 16 GB NVIDIA	16 GB	897 GB/s	Jun 2017	14.3 GB	IQ4_XS	Tight
37.4 tok/s 22–60 · low confidence	GeForce RTX 5070 Ti NVIDIA	16 GB	896 GB/s	Feb 2025	14.3 GB	IQ4_XS	Tight

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

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: Rhymes AI
Organisation type: Industry
Country: United States of America
Published: 8 October 2024
Authors: Dongxu Li, Yudong Liu, Haoning Wu, Yue Wang, Zhiqi Shen, Bowen Qu, Xinyao Niu, Fan Zhou, Chengen Huang, Yanpeng Li, Chongyan Zhu, Xiaoyi Ren, Chao Li, Yifan Ye, Peng Liu, Lihuan Zhang, Hanshu Yan, Guoyin Wang, Bei Chen, Junnan Li

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Multimodal, Language, Video, Vision
Task: Language modeling/generation, Visual question answering, Image captioning, Question answering, Code generation, Video description, Character recognition (OCR)

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 24.9B
Training data: 6,800,000,000,000 tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 1.4 × 10²³ FLOP
How it was established: Operation counting

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (unrestricted)
Training code: Unreleased
Hugging Face: rhymes-ai

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Record confidence: Confident
Citations: 130

Sources

Where this record came from and when it was last checked.

Reference: ARIA : An Open Multimodal Native Mixture-of-Experts Model
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Aria

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run Aria

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

Hardware requirements in practice

Minimum card

Xeon Phi 7120P

Memory needed

14.3 GB

Fastest

136 tok/s

Aria reaches a parameter count of 24.9B. That lands in the range a serious desktop card can handle once the weights are compressed. The number of cards we track that can run it: 241.

The entry point is Xeon Phi 7120P, with a memory capacity of 16 GB, running it at a compression of IQ4_XS and producing around 9.6 tokens per second.

The quickest result comes from B200, generating roughly 136 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Where it came from

Aria was published by Rhymes AI, in the country recorded as United States of America, during October 2024. The publishing organisation is categorised as industry.

It works in the domain of Multimodal, Language, Video, Vision, and is recorded as performing the task of language modeling/generation, Visual question answering, Image captioning, Question answering, Code generation, Video description, Character recognition (OCR).

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached. On Hugging Face it is published under the organisation rhymes-ai.

Understanding the speeds

Half the cards that hold it manage more than 18.7 tokens per second. Clearing the ten tokens per second that roughly matches reading speed: 191 of them.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

Without the attention layout on record, the memory column is an approximation. It is close enough to choose hardware by, and least reliable at long context.

What went into building it

Producing it required arithmetic totalling around 1.4 × 10²³ FLOP. That figure measures what producing the model cost, and has no bearing on how fast it answers.

It was trained on a corpus of about 6,800,000,000,000 tokens of text.

Step by step

How to choose a GPU for Aria

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Start from the memory column

Start from what it actually needs, which is the requirement of Aria, needing around 14.3 GB at a compression of IQ4_XS. That figure, not the headline performance of a card, is what decides whether it runs.
02

Decide how long your conversations run

The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect, because at long context a card that handles short questions easily can be dropped by Aria.
03

Choose how far you will compress it

The quantisation column varies by card, because a bigger card holds a more accurate copy, reaching a compression of IQ4_XS on the smallest card that fits. Setting a minimum quality drops the cards that only manage it by squeezing further than you would want, and holds the comparison at one level.
04

Compare tokens per second, not specifications

The speed ordering is effectively an ordering by memory bandwidth, for Aria. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 136 tok/s.
05

Check the fit verdict before buying

The fit column separates cards that just manage it from those with room to spare, in the case of Aria. Comfortable means you can grow the context later. That difference matters more than a few tokens per second, so buy for comfortable if you expect to.
06

Open the card you have settled on

Every card name links to its own page, which runs the same calculation across the whole model catalogue. A card is usually bought for more than one model, so it is worth a look before buying for Aria.

Answers

Aria — common questions

Aria— what GPU do I need to run it?

The smallest card in our catalogue that holds it is Xeon Phi 7120P, with a memory capacity of 16 GB. It runs the model at a compression of IQ4_XS using about 14.3 GB, and produces roughly 9.6 tokens per second. The number of cards able to run it in total: 241.

Aria— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 136 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and the number of cards clearing that: 191.

Aria— how much VRAM does it need?

It needs about 14.3 GB at a compression of IQ4_XS, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Aria— can I run it on a GPU holding 16 GB?

Yes. The card Tesla V100 SXM2 16 GB, holding 16 GB, runs it at a compression of IQ4_XS, using about 14.3 GB and generating roughly 47.2 tokens per second. The fit is tight.

Aria— can I run it on a GPU holding 24 GB?

Yes. The card GeForce RTX 5090 D V2, holding 24 GB, runs it at a compression of Q6_K, using about 21.6 GB and generating roughly 33.1 tokens per second. The fit is tight.

Aria— is it open source?

Its weights are published, so it can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

Aria— how many parameters does it have?

It has a parameter count of 24.9B. ARIA MoE activates 3.5B parameters per text token and has a total of 24.9B parameter. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Aria— who created it?

It was published by Rhymes AI, based in United States of America, an organisation categorised as industry.

Aria— when was it released?

It was published in October 2024. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Aria— what is it used for?

It works in the domain of Multimodal, Language, Video, Vision, and is recorded as handling the task of language modeling/generation, Visual question answering, Image captioning, Question answering, Code generation, Video description, Character recognition (OCR). Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

Aria— where can I download it?

Its weights are published on Hugging Face, under the organisation rhymes-ai. We do not host model files — this site calculates what hardware is needed to run them.

Aria— how much compute was used to train it?

Training consumed around 1.4 × 10²³ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

Aria— can I run it if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part held in system memory drags the whole thing down. The nearest miss we calculate falls short by 5.0 GB. Every figure here assumes the whole model is resident on the card.

Aria— would two GPUs run it faster?

A second card roughly doubles the memory available but not the generation rate. The number already able to run it alone: 241. So a second card is rarely the answer here.

Aria— why does the quantisation differ between cards?

A larger card holds a more accurate copy. The number of compression levels used across the cards that run it: 4. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

Aria— how accurate are these speed estimates?

They are calculated from specifications rather than measured, and each carries a range. One example: 82–218 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

Source

Original publication

Record last updated 25 May 2026

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models