Flan T5-XXL + BLIP-2 TPS calculator

Open weights Salesforce Research 12.1B parameters January 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

509 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Xeon Phi 5110P

8 GB · Q3_K_M · 19.7 tok/s

Fastest card

B200

280 tok/s · 180 GB

Which GPUs can run Flan T5-XXL + BLIP-2?

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

509 cards match

Calculating

					Needs	Quantisation	Fit
280 tok/s 168–448 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	13.7 GB	Q8_0	Comfortable
280 tok/s 168–448 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	13.7 GB	Q8_0	Comfortable
224 tok/s 134–358 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	13.7 GB	Q8_0	Comfortable
224 tok/s 134–358 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	13.7 GB	Q8_0	Comfortable
179 tok/s 107–286 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	13.7 GB	Q8_0	Comfortable
171 tok/s 103–274 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	13.7 GB	Q8_0	Comfortable
171 tok/s 103–274 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	13.7 GB	Q8_0	Comfortable
164 tok/s 98–262 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	13.7 GB	Q8_0	Comfortable
145 tok/s 87–233 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	13.7 GB	Q8_0	Comfortable
145 tok/s 87–233 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	13.7 GB	Q8_0	Comfortable
145 tok/s 87–233 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	13.7 GB	Q8_0	Comfortable
141 tok/s 84–225 · low confidence	CMP 170HX 8 GB NVIDIA	8 GB	1,490 GB/s	Sep 2021	6.6 GB	Q3_K_M	Tight
138 tok/s 83–221 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	13.7 GB	Q8_0	Comfortable
126 tok/s 76–202 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.0 GB	Q4_K_M	Tight
118 tok/s 71–188 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	13.7 GB	Q8_0	Comfortable
118 tok/s 71–188 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	13.7 GB	Q8_0	Comfortable
118 tok/s 71–188 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	13.7 GB	Q8_0	Comfortable
118 tok/s 71–188 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	13.7 GB	Q8_0	Comfortable
118 tok/s 71–188 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	13.7 GB	Q8_0	Comfortable
89.6 tok/s 54–143 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	13.7 GB	Q8_0	Comfortable
89.6 tok/s 54–143 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	13.7 GB	Q8_0	Comfortable
74.6 tok/s 45–119 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	13.7 GB	Q8_0	Comfortable
73.0 tok/s 44–117 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	13.7 GB	Q8_0	Comfortable
72.5 tok/s 44–116 · low confidence	RTX A5000-8Q NVIDIA	8 GB	768 GB/s	Apr 2021	6.6 GB	Q3_K_M	Tight
71.4 tok/s 43–114 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	13.7 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: Salesforce Research
Organisation type: Industry
Country: United States of America
Published: 30 January 2023
Authors: Junnan Li, Dongxu Li, Silvio Savarese, Steven Hoi

What it does

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

Domain: Multimodal, Language, Vision
Task: Vision-language generation, Chat, Visual question answering
Base model: Flan-T5 11B
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: 12.1B
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.

Fine-tuning compute: 1.2 × 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 40 GB
Wall-clock time: 200 hours (8.3 days)
Compute cost: $99,690

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 (unrestricted)
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.

Record confidence: Confident
Citations: 8,019

Sources

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

Reference: BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Flan T5-XXL + BLIP-2

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 Flan T5-XXL + BLIP-2

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

Xeon Phi 5110P

Memory needed

6.6 GB

Fastest

280 tok/s

Flan T5-XXL + BLIP-2 reaches a parameter count of 12.1B. 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: 509.

The smallest card that holds it is Xeon Phi 5110P, with a memory capacity of 8 GB, running it at a compression of Q3_K_M and producing around 19.7 tokens per second.

At the other end sits B200, generating roughly 280 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Where it came from

Flan T5-XXL + BLIP-2 was published by Salesforce Research, in the country recorded as United States of America, during January 2023. The category the publisher falls under is industry.

It works in the domain of Multimodal, Language, Vision, and is recorded as performing the task of vision-language generation, Chat, Visual question answering.

Its starting point was an existing base model, Flan-T5 11B. That is the usual way a specialised model is produced.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached.

Understanding the speeds

The median result is around 21.1 tokens per second. Exceeding reading speed outright: 455 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.

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.

Step by step

How to choose a GPU for Flan T5-XXL + BLIP-2

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 able to hold Flan T5-XXL + BLIP-2, needing around 6.6 GB at a compression of Q3_K_M. Capacity is the gate — a card either holds it or it does not.
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 Flan T5-XXL + BLIP-2.
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

Sort by speed

The speed ordering is effectively an ordering by memory bandwidth, for Flan T5-XXL + BLIP-2. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 280 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 Flan T5-XXL + BLIP-2. 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

Each card page repeats this sweep for every model we hold, answering what else the hardware is good for beyond Flan T5-XXL + BLIP-2.

Answers

Flan T5-XXL + BLIP-2 — common questions

Flan T5-XXL + BLIP-2— would two GPUs run it faster?

Capacity adds across cards; throughput does not. The number of cards already holding it on their own: 509. So a second card is rarely the answer here.

Flan T5-XXL + BLIP-2— why does the quantisation differ between cards?

Because capacity varies, so does how hard it has to be squeezed. The number of distinct levels in the table above: 4. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

Flan T5-XXL + BLIP-2— 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: 168–448 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

Flan T5-XXL + BLIP-2— what GPU do I need to run it?

The smallest card in our catalogue that holds it is Xeon Phi 5110P, with a memory capacity of 8 GB. It runs the model at a compression of Q3_K_M using about 6.6 GB, and produces roughly 19.7 tokens per second. The number of cards able to run it in total: 509.

Flan T5-XXL + BLIP-2— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 280 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: 455.

Flan T5-XXL + BLIP-2— how much VRAM does it need?

It needs about 6.6 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.

Flan T5-XXL + BLIP-2— can I run it on a GPU holding 8 GB?

Yes. The card CMP 170HX 8 GB, holding 8 GB, runs it at a compression of Q3_K_M, using about 6.6 GB and generating roughly 141 tokens per second. The fit is tight.

Flan T5-XXL + BLIP-2— 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 9.4 GB and generating roughly 57.0 tokens per second. The fit is tight.

Flan T5-XXL + BLIP-2— 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 Q8_0, using about 13.7 GB and generating roughly 39.6 tokens per second. The fit is tight.

Flan T5-XXL + BLIP-2— 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 13.7 GB and generating roughly 46.9 tokens per second. The fit is comfortable.

Flan T5-XXL + BLIP-2— 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.

Flan T5-XXL + BLIP-2— how many parameters does it have?

It has a parameter count of 12.1B. 12.1B, per Table 2. only 108M trainable params (i.e. params trained during the BLIP process). 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.

Flan T5-XXL + BLIP-2— who created it?

It was published by Salesforce Research, based in United States of America, an organisation categorised as industry.

Flan T5-XXL + BLIP-2— when was it released?

It was published in January 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.

Flan T5-XXL + BLIP-2— what is it used for?

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

Flan T5-XXL + BLIP-2— 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.

Flan T5-XXL + BLIP-2— 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.

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