VILA-13B TPS calculator

Open weights NVIDIA,Massachusetts Institute of Technology (MIT) 13.4B 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

306 cards that can run it

818 cards we hold specifications for

Smallest card that fits

P102-101

10 GB · Q4_K_M · 19.9 tok/s

Fastest card

B200

254 tok/s · 180 GB

Which GPUs can run VILA-13B?

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
254 tok/s 152–406 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	15.0 GB	Q8_0	Comfortable
254 tok/s 152–406 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	15.0 GB	Q8_0	Comfortable
203 tok/s 122–324 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	15.0 GB	Q8_0	Comfortable
203 tok/s 122–324 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	15.0 GB	Q8_0	Comfortable
162 tok/s 97–259 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	15.0 GB	Q8_0	Comfortable
155 tok/s 93–248 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	15.0 GB	Q8_0	Comfortable
155 tok/s 93–248 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	15.0 GB	Q8_0	Comfortable
148 tok/s 89–238 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	15.0 GB	Q8_0	Comfortable
132 tok/s 79–211 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	15.0 GB	Q8_0	Comfortable
132 tok/s 79–211 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	15.0 GB	Q8_0	Comfortable
132 tok/s 79–211 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	15.0 GB	Q8_0	Comfortable
125 tok/s 75–200 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	15.0 GB	Q8_0	Comfortable
114 tok/s 69–183 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.8 GB	Q4_K_M	Tight
107 tok/s 64–171 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	15.0 GB	Q8_0	Comfortable
107 tok/s 64–171 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	15.0 GB	Q8_0	Comfortable
107 tok/s 64–171 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	15.0 GB	Q8_0	Comfortable
107 tok/s 64–171 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	15.0 GB	Q8_0	Comfortable
107 tok/s 64–171 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	15.0 GB	Q8_0	Comfortable
81.2 tok/s 49–130 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	15.0 GB	Q8_0	Comfortable
81.2 tok/s 49–130 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	15.0 GB	Q8_0	Comfortable
67.6 tok/s 41–108 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	15.0 GB	Q8_0	Comfortable
66.2 tok/s 40–106 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	15.0 GB	Q8_0	Comfortable
64.7 tok/s 39–104 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	15.0 GB	Q8_0	Comfortable
64.7 tok/s 39–104 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	15.0 GB	Q8_0	Comfortable
64.7 tok/s 39–104 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	15.0 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: NVIDIA,Massachusetts Institute of Technology (MIT)
Organisation type: Industry,Academia
Country: United States of America
Published: 12 December 2023
Authors: Ji Lin, Hongxu Yin, Wei Ping, Yao Lu, Pavlo Molchanov, Andrew Tao, Huizi Mao, Jan Kautz, Mohammad Shoeybi, Song Han

What it does

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

Domain: Multimodal, Language, Vision
Task: Chat, Visual question answering, Image captioning, Language modeling/generation, Question answering
Base model: Llama 2-13B,CLIP (ViT L/14@336px)
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: 13.4B
Training data: 32,430,000,000 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: 2.3 × 10²¹ FLOP

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 SXM4 80 GB
Chips used: 128
Power draw: 101.5 kW

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 (non-commercial)
Training code: Open source
Hugging Face: Efficient-Large-Model

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: Confident
Citations: 827

Sources

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

Reference: VILA: On Pre-training for Visual Language Models
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run VILA-13B

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 VILA-13B

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

P102-101

Memory needed

8.8 GB

Fastest

254 tok/s

VILA-13B is small enough at 13.4B parameters that hardware is rarely the obstacle — 306 of the cards we track can run it, including cards several years old.

The entry point is the P102-101: 10 GB of memory, Q4_K_M compression, roughly 19.9 tokens per second.

Top of the range is the B200, at roughly 254 tokens per second thanks to 8,000 GB/s of bandwidth.

Background

VILA-13B was published by NVIDIA,Massachusetts Institute of Technology (MIT), in United States of America, in December 2023. The organisation is categorised as industry,Academia.

It works in Multimodal, Language, Vision, and is recorded as doing chat, Visual question answering, Image captioning, Language modeling/generation, Question answering.

Its starting point was Llama 2-13B,CLIP (ViT L/14@336px) — most models at this scale are adapted from an existing base rather than built from nothing.

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

Reading the throughput figures

The median result is around 21.4 tokens per second; 269 cards produce text faster than most people read it.

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.

How it was trained

Training it took roughly 2.3 × 10²¹ FLOP of computation, on NVIDIA A100 SXM4 80 GB — a measure of what producing the model cost, not of how fast it answers.

The training set ran to roughly 32,430,000,000 tokens.

Its inclusion criterion is sOTA improvement.

Step by step

How to choose a GPU for VILA-13B

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

01

Check what it needs before anything else

The table lists every card that can hold VILA-13B — around 8.8 GB at Q4_K_M. That figure, not the card's headline performance, is what decides whether it runs.
02

Match the context to your actual use

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for VILA-13B.
03

Choose how far you will compress it

Compression is what makes VILA-13B fit smaller cards, at some cost in accuracy — Q4_K_M on the smallest card that fits. A minimum quality removes the ones that go too far.
04

Sort by speed

The speed ordering for VILA-13B is effectively an ordering by memory bandwidth, which is why the B200 tops it at 254 tok/s.
05

Check the fit verdict before buying

The fit column separates cards that just manage VILA-13B from those with room to spare. Buy for the second if the context might grow.
06

Check the card from the other side

Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond VILA-13B.

Answers

VILA-13B — common questions

What GPU do I need to run VILA-13B?

The smallest card in our catalogue that holds VILA-13B is the P102-101, with 10 GB of memory. It runs the model at Q4_K_M using about 8.8 GB, and produces roughly 19.9 tokens per second. 306 cards in total can run it.

How fast is VILA-13B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 254 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 269 of the cards that can run VILA-13B clear that.

How much VRAM does VILA-13B need?

About 8.8 GB at Q4_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 VILA-13B on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q5_K_M, using about 10.3 GB and generating roughly 51.7 tokens per second — a tight fit.

Can I run VILA-13B on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q6_K, using about 11.9 GB and generating roughly 52.1 tokens per second — a comfortable fit.

Can I run VILA-13B on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 15.0 GB and generating roughly 42.5 tokens per second — a comfortable fit.

Is VILA-13B open source?

Its weights are published, so VILA-13B 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 VILA-13B have?

VILA-13B has 13.4B parameters. https://huggingface.co/Efficient-Large-Model/VILA-13b/tree/main?show_file_info=model.safetensors.index.json. 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 VILA-13B?

VILA-13B was published by NVIDIA,Massachusetts Institute of Technology (MIT), based in United States of America, categorised as industry,Academia.

When was VILA-13B released?

VILA-13B 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.

What is VILA-13B used for?

VILA-13B works in Multimodal, Language, Vision, and is recorded as handling chat, Visual question answering, Image captioning, Language modeling/generation, Question answering. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download VILA-13B?

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

How much compute was used to train VILA-13B?

Around 2.3 × 10²¹ FLOP, on NVIDIA A100 SXM4 80 GB. 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.

Can I run VILA-13B if it does not fit in my GPU?

It can be split between the card and system memory, but VILA-13B generates painfully slowly that way — the nearest miss we calculate is short by 1.6 GB. Nothing on this page assumes offloading.

Would two GPUs run VILA-13B faster?

Two cards buy memory rather than speed. That matters for VILA-13B only if one card cannot hold it — 306 can, so a second adds little.

Why does the quantisation differ between cards for VILA-13B?

A larger card holds a more accurate copy. Across the cards that run VILA-13B, 4 compression levels are used; the floor control above pins it to one.

How accurate are these VILA-13B speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 152–406 tok/s on the B200, for instance. 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