Hunyuan-Large TPS calculator

Open weights Tencent 389B parameters November 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

4 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Radeon Instinct MI325X

256 GB · IQ4_XS · 12.5 tok/s

Fastest card

B300

20.1 tok/s · 288 GB

Which GPUs can run Hunyuan-Large?

Set the inputs, read the answer

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

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

4 cards match

Calculating
Needs Quantisation Fit
20.1 tok/s

12–32 · low confidence

B300 NVIDIA 288 GB 8,000 GB/s Sep 2025 236.0 GB Q4_K_M Tight
16.1 tok/s

10–26 · low confidence

Radeon Instinct MI350X AMD 288 GB 8,190 GB/s Jan 2025 236.0 GB Q4_K_M Tight
16.1 tok/s

10–26 · low confidence

Radeon Instinct MI355X AMD 288 GB 8,190 GB/s Jan 2025 236.0 GB Q4_K_M Tight
12.5 tok/s

8–20 · low confidence

Radeon Instinct MI325X AMD 256 GB 6,000 GB/s Oct 2024 213.4 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
Tencent
Organisation type
Industry
Country
China
Published
6 November 2024
Authors
Xingwu Sun, Yanfeng Chen, Yiqing Huang, Ruobing Xie, Jiaqi Zhu, Kai Zhang, Shuaipeng Li, Zhen Yang, Jonny Han, Xiaobo Shu, Jiahao Bu, Zhongzhi Chen, Xuemeng Huang, Fengzong Lian, Saiyong Yang, Jianfeng Yan, Yuyuan Zeng, Xiaoqin Ren, Chao Yu, Lulu Wu, Yue Mao, Jun Xia, Tao Yang, Suncong Zheng, Kan Wu, Dian Jiao, Jinbao Xue, Xipeng Zhang, Decheng Wu, Kai Liu, Dengpeng Wu, Guanghui Xu, Shaohua Chen, …

What it does

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

Domain
Language
Task
Language modeling/generation, Question answering, Code generation, Translation
Numerical format
BF16

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
389B

"a total of 389 billion parameters and 52 billion activation parameters"

Training data
7,000,000,000,000 tokens

"# Trained Tokens 7T" Table 1

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
3.5 × 10²⁴ FLOP

52B activated parameters 6ND = 6*52*10^9*7*10^12 = 2.184 × 10^24 They also suggest more precise formula to calculate MoE compute budget: 9.59ND + 2.3 × 10^8D = 9.59*52*10^9*7*10^12 + 2.3 × 10^8 × 7*10^12 = 3.49237×10^24 which seems closer to projected compute on Figure 3

How it was established
Operation counting

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Data centre
The paper does not mention any hardware, GPUs or any information regarding the hardware used.

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 (restricted use)
Training code
Open (restricted use)

the license doesn't regulate usage in the EU also requires additional licensing in case of massive commercial use

How it is classified

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

Likely above 10²³ FLOP
Yes
Why it is tracked
Training cost
Record confidence
Confident

Sources

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

Reference
Hunyuan-Large: An Open-Source MoE Model with 52 Billion Activated Parameters by Tencent
Last updated
18 December 2025

The extremes

The ten fastest GPUs that run Hunyuan-Large

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.

  1. 01 B300 288 GB · 8,000 GB/s · Q4_K_M 20.1 tok/s
  2. 02 Radeon Instinct MI350X 288 GB · 8,190 GB/s · Q4_K_M 16.1 tok/s
  3. 03 Radeon Instinct MI355X 288 GB · 8,190 GB/s · Q4_K_M 16.1 tok/s
  4. 04 Radeon Instinct MI325X 256 GB · 6,000 GB/s · IQ4_XS 12.5 tok/s

What the numbers mean

What it takes to run this model

Minimum card

Radeon Instinct MI325X

Memory needed

213.4 GB

Fastest

20.1 tok/s

At 389B parameters, Hunyuan-Large is beyond what any single graphics card holds. Running it means either splitting it across several cards or renting hardware built for the job — 4 of the cards we track can hold it on their own, and all of them are datacentre parts.

The smallest card that holds it is the Radeon Instinct MI325X with 256 GB, running it at IQ4_XS and producing around 12.5 tokens per second.

A B300 is the fastest we calculate for it: about 20.1 tokens per second, from 8,000 GB/s of memory bandwidth.

What this model is

Hunyuan-Large was published by Tencent, in China, in November 2024. The organisation is categorised as industry.

It works in Language, and is recorded as doing language modeling/generation, Question answering, Code generation, Translation.

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all.

What decides the speed

Half the cards that hold it manage more than 16.1 tokens per second, and 4 exceed reading speed outright.

Being dense, it reads all of itself per token, which is why the ordering by speed below follows the ordering by memory bandwidth so closely.

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.

Training and provenance

Training it took roughly 3.5 × 10²⁴ FLOP of computation — a measure of what producing the model cost, not of how fast it answers.

It was trained on about 7,000,000,000,000 tokens of text.

Its inclusion criterion is training cost.

Step by step

How to choose a GPU for Hunyuan-Large

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

  1. 01

    Start from the memory column

    Look at what Hunyuan-Large actually needs — around 213.4 GB at IQ4_XS. No amount of processing power compensates for a card that cannot hold it.

  2. 02

    Decide how long your conversations run

    Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for Hunyuan-Large.

  3. 03

    Decide how much compression you will accept

    Each card runs the least-compressed copy it can hold — IQ4_XS on the smallest card that fits. Setting a floor drops the cards that only manage Hunyuan-Large by squeezing it further than you would want.

  4. 04

    Compare tokens per second, not specifications

    Ranking by tokens per second for Hunyuan-Large follows memory bandwidth, not core counts, which is why the B300 tops it at 20.1 tok/s.

  5. 05

    Check the fit verdict before buying

    Tight means Hunyuan-Large loads and works, with no room to raise the context later. Comfortable means you can. The difference matters more than a few tokens per second.

  6. 06

    Check the card from the other side

    Following a card through to its own page shows every other model it can hold, which is the question that follows once Hunyuan-Large is settled.

Answers

Hunyuan-Large — common questions

01

Who created Hunyuan-Large?

Hunyuan-Large was published by Tencent, based in China, categorised as industry.

02

When was Hunyuan-Large released?

Hunyuan-Large was published in November 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.

03

What is Hunyuan-Large used for?

Hunyuan-Large works in Language, and is recorded as handling language modeling/generation, Question answering, Code generation, Translation. These are the areas it was designed around; they describe intent rather than a hard boundary.

04

Where can I download Hunyuan-Large?

The weights for Hunyuan-Large are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

05

How much compute was used to train Hunyuan-Large?

Around 3.5 × 10²⁴ 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.

06

Can I run Hunyuan-Large if it does not fit in my GPU?

It can be split between the card and system memory, but Hunyuan-Large generates painfully slowly that way — the nearest miss we calculate is short by 63.2 GB. Nothing on this page assumes offloading.

07

Would two GPUs run Hunyuan-Large faster?

A second card roughly doubles the memory available but not the generation rate. With 4 cards already able to run Hunyuan-Large alone, the case for pairing is weak.

08

Why does the quantisation differ between cards for Hunyuan-Large?

A larger card holds a more accurate copy. Across the cards that run Hunyuan-Large, 2 compression levels are used; the floor control above pins it to one.

09

How accurate are these Hunyuan-Large speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 12–32 tok/s on the B300, for instance. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

10

What GPU do I need to run Hunyuan-Large?

The smallest card in our catalogue that holds Hunyuan-Large is the Radeon Instinct MI325X, with 256 GB of memory. It runs the model at IQ4_XS using about 213.4 GB, and produces roughly 12.5 tokens per second. 4 cards in total can run it.

11

How fast is Hunyuan-Large on a GPU?

It depends on the card. The quickest we calculate is a B300 at about 20.1 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 4 of the cards that can run Hunyuan-Large clear that.

12

How much VRAM does Hunyuan-Large need?

About 213.4 GB at IQ4_XS compression, 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.

13

Is Hunyuan-Large open source?

Its weights are published, so Hunyuan-Large 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.

14

How many parameters does Hunyuan-Large have?

Hunyuan-Large has 389B parameters. "a total of 389 billion parameters and 52 billion activation parameters". 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.

Source

Original publication

Record last updated 18 December 2025

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.