Qwen-VL TPS calculator

Open weights Alibaba 9.6B parameters August 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

582 of 818 cards that can run it

Smallest card that fits

Quadro 6000

6 GB · Q3_K_M · 14.5 tok/s

Fastest card

B200

353 tok/s · 180 GB

Which GPUs can run Qwen-VL?

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

582 cards match

Calculating

					Needs	Quantisation	Fit
353 tok/s 212–565 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	11.0 GB	Q8_0	Comfortable
353 tok/s 212–565 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	11.0 GB	Q8_0	Comfortable
282 tok/s 169–451 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	11.0 GB	Q8_0	Comfortable
282 tok/s 169–451 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	11.0 GB	Q8_0	Comfortable
225 tok/s 135–361 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	11.0 GB	Q8_0	Comfortable
216 tok/s 129–345 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	11.0 GB	Q8_0	Comfortable
216 tok/s 129–345 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	11.0 GB	Q8_0	Comfortable
206 tok/s 124–330 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	11.0 GB	Q8_0	Comfortable
183 tok/s 110–293 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	11.0 GB	Q8_0	Comfortable
183 tok/s 110–293 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	11.0 GB	Q8_0	Comfortable
183 tok/s 110–293 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	11.0 GB	Q8_0	Comfortable
174 tok/s 104–278 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	11.0 GB	Q8_0	Comfortable
152 tok/s 91–243 · low confidence	CMP 170HX 8 GB NVIDIA	8 GB	1,490 GB/s	Sep 2021	6.5 GB	Q4_K_M	Tight
148 tok/s 89–237 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	11.0 GB	Q8_0	Comfortable
148 tok/s 89–237 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	11.0 GB	Q8_0	Comfortable
148 tok/s 89–237 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	11.0 GB	Q8_0	Comfortable
148 tok/s 89–237 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	11.0 GB	Q8_0	Comfortable
148 tok/s 89–237 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	11.0 GB	Q8_0	Comfortable
113 tok/s 68–181 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	11.0 GB	Q8_0	Comfortable
113 tok/s 68–181 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	11.0 GB	Q8_0	Comfortable
100 tok/s 60–160 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.7 GB	Q6_K	Tight
94.1 tok/s 56–150 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	11.0 GB	Q8_0	Comfortable
92.1 tok/s 55–147 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	11.0 GB	Q8_0	Comfortable
90.0 tok/s 54–144 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	11.0 GB	Q8_0	Comfortable
90.0 tok/s 54–144 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	11.0 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: Alibaba
Organisation type: Industry
Country: China
Published: 24 August 2023
Authors: Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Junyang Lin, Chang Zhou, Jingren Zhou

What it does

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

Domain: Multimodal, Language, Vision
Task: Image captioning, Chat, Question answering, Visual question answering
Base model: Qwen-7B
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: 9.6B
Training data: 500,000,000,000 tokens
Epochs: 1

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.

Why it is tracked: SOTA improvement
Record confidence: Likely
Citations: 1,998

Sources

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

Reference: Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond
Last updated: 25 May 2026

The extremes

The ten fastest GPUs for Qwen-VL

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 Qwen-VL

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

Quadro 6000

Memory needed

5.4 GB

Fastest

353 tok/s

Qwen-VL is small enough at 9.6B parameters that hardware is rarely the obstacle — 582 of the cards we track can run it, including cards several years old.

The smallest card that holds it is the Quadro 6000 with 6 GB, running it at Q3_K_M and producing around 14.5 tokens per second.

Top of the range is the B200, at roughly 353 tokens per second thanks to 8,000 GB/s of bandwidth.

About this model

Qwen-VL was published by Alibaba, in China, in August 2023. The organisation is categorised as industry.

It works in Multimodal, Language, Vision, and is recorded as doing image captioning, Chat, Question answering, Visual question answering.

It is derived from Qwen-7B rather than trained from scratch, which 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.

How fast it runs, and why

Across every card that can run it, the middle of the range is about 22.9 tokens per second, and 544 of them clear the ten tokens per second that roughly matches reading speed.

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.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

Training and provenance

The training set ran to roughly 500,000,000,000 tokens.

Its inclusion criterion is sOTA improvement.

Step by step

How to choose a GPU for Qwen-VL

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 Qwen-VL — around 5.4 GB at Q3_K_M. Capacity is the gate — a card either holds it or it does not.
02

Set the context length you will work at

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for Qwen-VL.
03

Decide how much compression you will accept

Each card runs the least-compressed copy it can hold — Q3_K_M on the smallest card that fits. Setting a floor drops the cards that only manage Qwen-VL by squeezing it further than you would want.
04

Compare tokens per second, not specifications

The speed ordering for Qwen-VL is effectively an ordering by memory bandwidth, which is why the B200 tops it at 353 tok/s.
05

Check the fit verdict before buying

A tight fit runs Qwen-VL but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
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 Qwen-VL is settled.

Answers

Qwen-VL — common questions

Why does the quantisation differ between cards for Qwen-VL?

Each card is shown running the least-compressed copy it can hold, and Qwen-VL appears at 4 different compression levels across the cards that fit it. Bigger cards get the more accurate version.

How accurate are these Qwen-VL speed estimates?

These are estimates with real error bars. The fastest result here, 212–565 tok/s on the B200, could reasonably land anywhere in its published range depending on which runtime you use.

What GPU do I need to run Qwen-VL?

The smallest card in our catalogue that holds Qwen-VL is the Quadro 6000, with 6 GB of memory. It runs the model at Q3_K_M using about 5.4 GB, and produces roughly 14.5 tokens per second. 582 cards in total can run it.

How fast is Qwen-VL on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 353 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 544 of the cards that can run Qwen-VL clear that.

How much VRAM does Qwen-VL need?

About 5.4 GB at Q3_K_M 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.

Can I run Qwen-VL on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q4_K_M, using about 6.5 GB and generating roughly 152 tokens per second — a tight fit.

Can I run Qwen-VL on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q6_K, using about 8.7 GB and generating roughly 58.5 tokens per second — a comfortable fit.

Can I run Qwen-VL on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 11.0 GB and generating roughly 49.9 tokens per second — a comfortable fit.

Can I run Qwen-VL on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 11.0 GB and generating roughly 59.1 tokens per second — a comfortable fit.

Is Qwen-VL open source?

Its weights are published, so Qwen-VL 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.

How many parameters does Qwen-VL have?

Qwen-VL has 9.6B parameters. 9.6B total - Table 1. 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.

Who created Qwen-VL?

Qwen-VL was published by Alibaba, based in China, categorised as industry.

When was Qwen-VL released?

Qwen-VL was published in August 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.

What is Qwen-VL used for?

Qwen-VL works in Multimodal, Language, Vision, and is recorded as handling image captioning, Chat, Question answering, Visual question answering. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

Where can I download Qwen-VL?

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

Can I run Qwen-VL if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part in system memory drags the whole thing down — the nearest miss we calculate is short by 2.0 GB. Our figures for Qwen-VL assume it is fully resident.

Would two GPUs run Qwen-VL faster?

Two cards buy memory rather than speed. That matters for Qwen-VL only if one card cannot hold it — 582 can, so a second adds little.

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