HAMSTER VLM TPS calculator

Open weights NVIDIA,University of Washington,University of Southern California 13.5B parameters February 2025

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 · 19.7 tok/s

Fastest card

B200

251 tok/s · 180 GB

Which GPUs can run HAMSTER VLM?

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
251 tok/s 151–402 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	15.1 GB	Q8_0	Comfortable
251 tok/s 151–402 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	15.1 GB	Q8_0	Comfortable
201 tok/s 120–321 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	15.1 GB	Q8_0	Comfortable
201 tok/s 120–321 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	15.1 GB	Q8_0	Comfortable
160 tok/s 96–257 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	15.1 GB	Q8_0	Comfortable
153 tok/s 92–246 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	15.1 GB	Q8_0	Comfortable
153 tok/s 92–246 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	15.1 GB	Q8_0	Comfortable
147 tok/s 88–235 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	15.1 GB	Q8_0	Comfortable
130 tok/s 78–209 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	15.1 GB	Q8_0	Comfortable
130 tok/s 78–209 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	15.1 GB	Q8_0	Comfortable
130 tok/s 78–209 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	15.1 GB	Q8_0	Comfortable
124 tok/s 74–198 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	15.1 GB	Q8_0	Comfortable
113 tok/s 68–181 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.9 GB	Q4_K_M	Tight
105 tok/s 63–169 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	15.1 GB	Q8_0	Comfortable
105 tok/s 63–169 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	15.1 GB	Q8_0	Comfortable
105 tok/s 63–169 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	15.1 GB	Q8_0	Comfortable
105 tok/s 63–169 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	15.1 GB	Q8_0	Comfortable
105 tok/s 63–169 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	15.1 GB	Q8_0	Comfortable
80.3 tok/s 48–128 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	15.1 GB	Q8_0	Comfortable
80.3 tok/s 48–128 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	15.1 GB	Q8_0	Comfortable
66.9 tok/s 40–107 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	15.1 GB	Q8_0	Comfortable
65.5 tok/s 39–105 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	15.1 GB	Q8_0	Comfortable
64.0 tok/s 38–102 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	15.1 GB	Q8_0	Comfortable
64.0 tok/s 38–102 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	15.1 GB	Q8_0	Comfortable
64.0 tok/s 38–102 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	15.1 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,University of Washington,University of Southern California
Organisation type: Industry,Academia,Academia
Country: United States of America
Published: 8 February 2025
Authors: Yi Li, Yuquan Deng, Jesse Zhang, Joel Jang, Marius Memmel, Raymond Yu, Caelan Reed Garrett, Fabio Ramos, Dieter Fox, Anqi Li, Abhishek Gupta, Ankit Goyal

What it does

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

Domain: Robotics
Task: Robotic manipulation
Base model: VILA1.5-13B
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: 13.5B
Training data: tokens
Batch size: 256

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: 2.4 × 10²¹ FLOP
How it was established: Hardware
Fine-tuning compute: 1.1 × 10²⁰ 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: 8
Wall-clock time: 30 hours
Power draw: 6.3 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: Unreleased

How it is classified

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

Record confidence: Confident
Citations: 104

Sources

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

Reference: HAMSTER: Hierarchical Action Models For Open-World Robot Manipulation
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run HAMSTER VLM

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 HAMSTER VLM

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

P102-101

Memory needed

8.9 GB

Fastest

251 tok/s

HAMSTER VLM reaches a parameter count of 13.5B. 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 smallest card that holds it is P102-101, with a memory capacity of 10 GB, running it at a compression of Q4_K_M and producing around 19.7 tokens per second.

The quickest result comes from B200, generating roughly 251 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Where it came from

HAMSTER VLM was published by NVIDIA,University of Washington,University of Southern California, in the country recorded as United States of America, during February 2025. It comes out of an organisation categorised as industry,Academia,Academia.

It works in the domain of Robotics, and is recorded as performing the task of robotic manipulation.

It builds on VILA1.5-13B. Most models at this scale are adapted from an existing base rather than built from nothing.

Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded.

Understanding the speeds

The median result is around 21.1 tokens per second. Exceeding reading speed outright: 269 of them.

Every weight participates in every token here, so bandwidth is the whole story: the ranking below is effectively a ranking of memory throughput.

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.

What went into building it

Producing it required arithmetic totalling around 2.4 × 10²¹ FLOP, on hardware recorded as NVIDIA A100 SXM4 80 GB. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Step by step

How to choose a GPU for HAMSTER VLM

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

The table lists every card able to hold HAMSTER VLM, needing around 8.9 GB at a compression of Q4_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

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

Decide how much compression you will accept

Each card runs the least-compressed copy it can hold, reaching a compression of Q4_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

Sort by speed to see how cards rank for HAMSTER VLM. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 251 tok/s.
05

Look at the headroom, not just the fit

A tight fit runs, but leaves nothing spare for a longer conversation, in the case of HAMSTER VLM. 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 HAMSTER VLM.

Answers

HAMSTER VLM — common questions

HAMSTER VLM— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 251 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: 269.

HAMSTER VLM— how much VRAM does it need?

It needs about 8.9 GB at a compression of Q4_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.

HAMSTER VLM— 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 Q5_K_M, using about 10.4 GB and generating roughly 51.1 tokens per second. The fit is tight.

HAMSTER VLM— 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 12.0 GB and generating roughly 51.5 tokens per second. The fit is tight.

HAMSTER VLM— 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 15.1 GB and generating roughly 42.1 tokens per second. The fit is comfortable.

HAMSTER VLM— 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.

HAMSTER VLM— how many parameters does it have?

It has a parameter count of 13.5B. https://huggingface.co/yili18/Hamster_dev VLM: VILA 1.5-13b, 13493916736 parameters policy: either RVT-2 or 3D-DA this record is for the HAMSTER VLM, so the policy is not included. 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.

HAMSTER VLM— who created it?

It was published by NVIDIA,University of Washington,University of Southern California, based in United States of America, an organisation categorised as industry,Academia,Academia.

HAMSTER VLM— when was it released?

It was published in February 2025.

HAMSTER VLM— what is it used for?

It works in the domain of Robotics, and is recorded as handling the task of robotic manipulation. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.

HAMSTER VLM— 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.

HAMSTER VLM— how much compute was used to train it?

Training consumed around 2.4 × 10²¹ FLOP, on hardware recorded as 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.

HAMSTER VLM— can I run it if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part held in system memory drags the whole thing down. The nearest miss we calculate falls short by 1.7 GB. Every figure here assumes the whole model is resident on the card.

HAMSTER VLM— would two GPUs run it faster?

Capacity adds across cards; throughput does not. The number of cards already holding it on their own: 306. So a second card is rarely the answer here.

HAMSTER VLM— 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: 4. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

HAMSTER VLM— 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: 151–402 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

HAMSTER VLM— 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 Q4_K_M using about 8.9 GB, and produces roughly 19.7 tokens per second. The number of cards able to run it in total: 306.

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