CogAgent TPS calculator

Open weights Tsinghua University,Z.ai (Zhipu AI) 18B parameters December 2023

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

295 cards that can run it

818 cards we hold specifications for

Smallest card that fits

GeForce GTX 1080 Ti

11 GB · Q3_K_M · 26.1 tok/s

Fastest card

B200

188 tok/s · 180 GB

Which GPUs can run CogAgent?

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

295 cards match

Calculating

					Needs	Quantisation	Fit
188 tok/s 113–301 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	20.0 GB	Q8_0	Comfortable
188 tok/s 113–301 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	20.0 GB	Q8_0	Comfortable
150 tok/s 90–241 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	20.0 GB	Q8_0	Comfortable
150 tok/s 90–241 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	20.0 GB	Q8_0	Comfortable
120 tok/s 72–192 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	20.0 GB	Q8_0	Comfortable
115 tok/s 69–184 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	20.0 GB	Q8_0	Comfortable
115 tok/s 69–184 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	20.0 GB	Q8_0	Comfortable
110 tok/s 66–176 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	20.0 GB	Q8_0	Comfortable
97.7 tok/s 59–156 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	20.0 GB	Q8_0	Comfortable
97.7 tok/s 59–156 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	20.0 GB	Q8_0	Comfortable
97.7 tok/s 59–156 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	20.0 GB	Q8_0	Comfortable
92.7 tok/s 56–148 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	20.0 GB	Q8_0	Comfortable
79.1 tok/s 47–126 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	20.0 GB	Q8_0	Comfortable
79.1 tok/s 47–126 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	20.0 GB	Q8_0	Comfortable
79.1 tok/s 47–126 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	20.0 GB	Q8_0	Comfortable
79.1 tok/s 47–126 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	20.0 GB	Q8_0	Comfortable
79.1 tok/s 47–126 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	20.0 GB	Q8_0	Comfortable
60.2 tok/s 36–96 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	20.0 GB	Q8_0	Comfortable
60.2 tok/s 36–96 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	20.0 GB	Q8_0	Comfortable
52.7 tok/s 32–84 · low confidence	GeForce RTX 3080 12 GB NVIDIA	12 GB	912 GB/s	Jan 2022	10.5 GB	IQ4_XS	Tight
52.7 tok/s 32–84 · low confidence	GeForce RTX 3080 Ti NVIDIA	12 GB	912 GB/s	May 2021	10.5 GB	IQ4_XS	Tight
50.2 tok/s 30–80 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	20.0 GB	Q8_0	Comfortable
49.1 tok/s 29–79 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	20.0 GB	Q8_0	Comfortable
48.0 tok/s 29–77 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	20.0 GB	Q8_0	Comfortable
48.0 tok/s 29–77 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	20.0 GB	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

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: Tsinghua University,Z.ai (Zhipu AI)
Organisation type: Academia,Industry
Country: China
Published: 14 December 2023
Authors: Wenyi Hong, Weihan Wang, Qingsong Lv, Jiazheng Xu, Wenmeng Yu, Junhui Ji, Yan Wang, Zihan Wang, Yuxuan Zhang, Juanzi Li, Bin Xu, Yuxiao Dong, Ming Ding, Jie Tang

What it does

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

Domain: Vision, Language
Task: Instruction interpretation, Visual question answering
Approach: Supervised
Base model: CogVLM-17B
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: 18B
Training data: 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: 6.7 × 10²² 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 (restricted use)
Training code: Open source

How it is classified

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

Why it is tracked: SOTA improvement
Record confidence: Likely
Citations: 731

Sources

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

Reference: CogAgent: A Visual Language Model for GUI Agents
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run CogAgent

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 CogAgent

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 it takes to run this model

Minimum card

GeForce GTX 1080 Ti

Memory needed

9.5 GB

Fastest

188 tok/s

CogAgent reaches a parameter count of 18B. 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: 295.

The smallest card that holds it is GeForce GTX 1080 Ti, with a memory capacity of 11 GB, running it at a compression of Q3_K_M and producing around 26.1 tokens per second.

At the other end sits B200, generating roughly 188 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Background

CogAgent was published by Tsinghua University,Z.ai (Zhipu AI), in the country recorded as China, during December 2023. The publishing organisation is categorised as academia,Industry.

It works in the domain of Vision, Language, and is recorded as performing the task of instruction interpretation, Visual question answering.

Its starting point was an existing base model, CogVLM-17B. That is the usual way a specialised model is produced.

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.

Reading the throughput figures

Half the cards that hold it manage more than 19.4 tokens per second. Exceeding reading speed outright: 248 of them.

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.

Its internal architecture is not on file, so memory is approximated from the parameter count and marked accordingly. Expect the real figure to differ, more so at long context.

What went into building it

Producing it required arithmetic totalling around 6.7 × 10²² FLOP. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Its inclusion criterion: sOTA improvement.

Step by step

How to choose a GPU for CogAgent

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

Every card here has been checked against CogAgent, needing around 9.5 GB at a compression of Q3_K_M. That figure, not the headline performance of a card, is what decides whether it runs.
02

Match the context to your actual use

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 CogAgent.
03

Decide how much compression you will accept

Compression is what makes a model fit smaller cards, at some cost in accuracy, 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 CogAgent. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 188 tok/s.
05

Look at the headroom, not just the fit

Tight means it loads and works with no room to raise the context later, in the case of CogAgent. 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

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 CogAgent.

Answers

CogAgent — common questions

CogAgent— how much compute was used to train it?

Training consumed around 6.7 × 10²² 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.

CogAgent— 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 2.6 GB. Every figure here assumes the whole model is resident on the card.

CogAgent— would two GPUs run it faster?

Capacity adds across cards; throughput does not. The number of cards already holding it on their own: 295. So a second card is rarely the answer here.

CogAgent— 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: 5. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

CogAgent— how accurate are these speed estimates?

These are estimates with real error bars, and any of them could reasonably land anywhere in its published range depending on which runtime you use. The fastest result here: 113–301 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

CogAgent— what GPU do I need to run it?

The smallest card in our catalogue that holds it is GeForce GTX 1080 Ti, with a memory capacity of 11 GB. It runs the model at a compression of Q3_K_M using about 9.5 GB, and produces roughly 26.1 tokens per second. The number of cards able to run it in total: 295.

CogAgent— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 188 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: 248.

CogAgent— how much VRAM does it need?

It needs about 9.5 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.

CogAgent— 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 IQ4_XS, using about 10.5 GB and generating roughly 52.7 tokens per second. The fit is tight.

CogAgent— 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 Q5_K_M, using about 13.7 GB and generating roughly 47.5 tokens per second. The fit is tight.

CogAgent— 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 20.0 GB and generating roughly 31.5 tokens per second. The fit is tight.

CogAgent— 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.

CogAgent— how many parameters does it have?

It has a parameter count of 18B. 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.

CogAgent— who created it?

It was published by Tsinghua University,Z.ai (Zhipu AI), based in China, an organisation categorised as academia,Industry.

CogAgent— when was it released?

It was published in December 2023. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

CogAgent— what is it used for?

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

CogAgent— 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.

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