DeepSeekMoE-16B TPS calculator

Open weights DeepSeek 16B parameters January 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

306 cards that can run it

818 cards we hold specifications for

Smallest card that fits

P102-101

10 GB · Q3_K_M · 108 tok/s

Fastest card

B200

1,176 tok/s · 180 GB

Which GPUs can run DeepSeekMoE-16B?

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

306 cards match

Calculating

					Needs	Quantisation	Fit
1,176 tok/s 706–1,882 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	17.5 GB	Q8_0	Comfortable
1,176 tok/s 706–1,882 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	17.5 GB	Q8_0	Comfortable
939 tok/s 564–1,503 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	17.5 GB	Q8_0	Comfortable
939 tok/s 564–1,503 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	17.5 GB	Q8_0	Comfortable
751 tok/s 451–1,202 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	17.5 GB	Q8_0	Comfortable
719 tok/s 431–1,151 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	17.5 GB	Q8_0	Comfortable
719 tok/s 431–1,151 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	17.5 GB	Q8_0	Comfortable
688 tok/s 413–1,101 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	17.5 GB	Q8_0	Comfortable
619 tok/s 371–990 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.1 GB	Q3_K_M	Tight
611 tok/s 366–977 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	17.5 GB	Q8_0	Comfortable
611 tok/s 366–977 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	17.5 GB	Q8_0	Comfortable
611 tok/s 366–977 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	17.5 GB	Q8_0	Comfortable
579 tok/s 348–927 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	17.5 GB	Q8_0	Comfortable
494 tok/s 296–791 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	17.5 GB	Q8_0	Comfortable
494 tok/s 296–791 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	17.5 GB	Q8_0	Comfortable
494 tok/s 296–791 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	17.5 GB	Q8_0	Comfortable
494 tok/s 296–791 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	17.5 GB	Q8_0	Comfortable
494 tok/s 296–791 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	17.5 GB	Q8_0	Comfortable
376 tok/s 226–602 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	17.5 GB	Q8_0	Comfortable
376 tok/s 226–602 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	17.5 GB	Q8_0	Comfortable
314 tok/s 188–502 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	17.5 GB	Q8_0	Comfortable
310 tok/s 186–496 · low confidence	GeForce RTX 3080 12 GB NVIDIA	12 GB	912 GB/s	Jan 2022	10.0 GB	Q4_K_M	Tight
310 tok/s 186–496 · low confidence	GeForce RTX 3080 Ti NVIDIA	12 GB	912 GB/s	May 2021	10.0 GB	Q4_K_M	Tight
307 tok/s 184–491 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	17.5 GB	Q8_0	Comfortable
302 tok/s 181–483 · low confidence	CMP 90HX NVIDIA	10 GB	760 GB/s	Jul 2021	8.1 GB	Q3_K_M	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: DeepSeek
Organisation type: Industry
Country: China
Published: 11 January 2024
Authors: Damai Dai, Chengqi Deng, Chenggang Zhao, R.X. Xu, Huazuo Gao, Deli Chen, Jiashi Li, Wangding Zeng, Xingkai Yu, Y. Wu, Zhenda Xie, Y.K. Li, Panpan Huang, Fuli Luo, Chong Ruan, Zhifang Sui, Wenfeng Liang

What it does

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

Domain: Language
Task: Chat

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: 16B
Training data: 2,000,000,000,000 tokens
Epochs: 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: 3.4 × 10²² FLOP
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.

Training hardware: NVIDIA A100,NVIDIA H800 SXM5

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: Unreleased

How it is classified

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

Record confidence: Likely
Citations: 844

Sources

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

Reference: DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run DeepSeekMoE-16B

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 DeepSeekMoE-16B

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

What you need to run it

Minimum card

P102-101

Memory needed

8.1 GB

Fastest

1,176 tok/s

DeepSeekMoE-16B reaches a parameter count of 16B. That is small enough that hardware is rarely the obstacle, including on cards several years old. The number of cards we track that can run it: 306.

The least hardware that works is P102-101, with a memory capacity of 10 GB, running it at a compression of Q3_K_M and producing around 108 tokens per second.

Top of the range is B200, generating roughly 1,176 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Where it came from

DeepSeekMoE-16B was published by DeepSeek, in the country recorded as China, during January 2024. The category the publisher falls under is industry.

It works in the domain of Language, and is recorded as performing the task of chat.

The weights being open is what puts this page in the calculator rather than only in the catalogue: it is a model you can actually hold.

Understanding the speeds

The median result is around 109.8 tokens per second. Producing text faster than most people read it: 304 of them.

Because it routes each token through a subset of its weights, it produces text at the pace of a much smaller model. The catch is memory: all of it still has to fit, so the speed is a bonus rather than a discount on hardware.

Its attention layout is on file, so the memory figures are computed exactly rather than approximated.

How it was trained

Training it took a computation budget of roughly 3.4 × 10²² FLOP, on hardware recorded as NVIDIA A100,NVIDIA H800 SXM5. 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 2,000,000,000,000 tokens of text.

Step by step

How to choose a GPU for DeepSeekMoE-16B

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

01

Read the memory figure first

Start from what it actually needs, which is the requirement of DeepSeekMoE-16B, needing around 8.1 GB at a compression of Q3_K_M. That figure, not the headline performance of a card, is what decides whether it runs.
02

Set the context length you will work at

Set the context to what you will actually use. The cache grows with the conversation, and it is the usual reason a card that seemed fine stops fitting DeepSeekMoE-16B.
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 Q3_K_M 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

Ranking by tokens per second follows memory bandwidth rather than core counts, for DeepSeekMoE-16B. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 1,176 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 DeepSeekMoE-16B. 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

See what else that card runs

Following a card through to its own page shows every other model it can hold, which is the question that follows once you have settled on DeepSeekMoE-16B.

Answers

DeepSeekMoE-16B — common questions

DeepSeekMoE-16B— when was it released?

It was published in January 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.

DeepSeekMoE-16B— what is it used for?

It works in the domain of Language, and is recorded as handling the task of chat. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

DeepSeekMoE-16B— where can I download it?

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

DeepSeekMoE-16B— how much compute was used to train it?

Training consumed around 3.4 × 10²² FLOP, on hardware recorded as NVIDIA A100,NVIDIA H800 SXM5. 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.

DeepSeekMoE-16B— can I run it if it does not fit in my GPU?

Partly. Layers that do not fit sit in system memory and run at a fraction of the speed, so a mostly-offloaded model is rarely worth using. The nearest miss we calculate falls short by 2.8 GB. Every figure here assumes the whole model is resident on the card.

DeepSeekMoE-16B— would two GPUs run it faster?

Two cards buy memory rather than speed, which matters only if one card cannot hold it. The number that can: 306. So a second card is rarely the answer here.

DeepSeekMoE-16B— why does the quantisation differ between cards?

Each card is shown running the least-compressed copy it can hold. The number of distinct compression levels across the cards that fit it: 5. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

DeepSeekMoE-16B— how accurate are these speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked, which is why each is published as a range rather than a single number. One example: 706–1,882 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

DeepSeekMoE-16B— what GPU do I need to run it?

The smallest card in our catalogue that holds it is P102-101, with a memory capacity of 10 GB. It runs the model at a compression of Q3_K_M using about 8.1 GB, and produces roughly 108 tokens per second. The number of cards able to run it in total: 306.

DeepSeekMoE-16B— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 1,176 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: 304.

DeepSeekMoE-16B— how much VRAM does it need?

It needs about 8.1 GB at a compression of Q3_K_M, 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.

DeepSeekMoE-16B— can I run it on a GPU holding 12 GB?

Yes. The card GeForce RTX 3080 Ti, holding 12 GB, runs it at a compression of Q4_K_M, using about 10.0 GB and generating roughly 310 tokens per second. The fit is tight.

DeepSeekMoE-16B— 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 Q6_K, using about 13.7 GB and generating roughly 241 tokens per second. The fit is tight.

DeepSeekMoE-16B— 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 Q8_0, using about 17.5 GB and generating roughly 197 tokens per second. The fit is comfortable.

DeepSeekMoE-16B— 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.

DeepSeekMoE-16B— how many parameters does it have?

It has a parameter count of 16B. 16B (total, but it's sparse). 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.

DeepSeekMoE-16B— who created it?

It was published by DeepSeek, based in China, an organisation categorised as industry.

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