Granite-Docling TPS calculator

Open weights IBM 258M parameters September 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

818 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q8_0 · 143 tok/s

Fastest card

B200

13,133 tok/s · 180 GB

Which GPUs can run Granite-Docling?

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

818 cards match

Calculating

					Needs	Quantisation	Fit
13,133 tok/s 7,880–21,012 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	1.0 GB	Q8_0	Comfortable
13,133 tok/s 7,880–21,012 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	1.0 GB	Q8_0	Comfortable
10,487 tok/s 6,292–16,779 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	1.0 GB	Q8_0	Comfortable
10,487 tok/s 6,292–16,779 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	1.0 GB	Q8_0	Comfortable
8,387 tok/s 5,032–13,419 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	1.0 GB	Q8_0	Comfortable
8,027 tok/s 4,816–12,844 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.0 GB	Q8_0	Comfortable
8,027 tok/s 4,816–12,844 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.0 GB	Q8_0	Comfortable
7,683 tok/s 4,610–12,292 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	1.0 GB	Q8_0	Comfortable
6,818 tok/s 4,091–10,909 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	1.0 GB	Q8_0	Comfortable
6,818 tok/s 4,091–10,909 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	1.0 GB	Q8_0	Comfortable
6,818 tok/s 4,091–10,909 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	1.0 GB	Q8_0	Comfortable
6,468 tok/s 3,881–10,349 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	1.0 GB	Q8_0	Comfortable
5,516 tok/s 3,309–8,825 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.0 GB	Q8_0	Comfortable
5,516 tok/s 3,309–8,825 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	1.0 GB	Q8_0	Comfortable
5,516 tok/s 3,309–8,825 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	1.0 GB	Q8_0	Comfortable
5,516 tok/s 3,309–8,825 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.0 GB	Q8_0	Comfortable
5,516 tok/s 3,309–8,825 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	1.0 GB	Q8_0	Comfortable
4,200 tok/s 2,520–6,720 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	1.0 GB	Q8_0	Comfortable
4,200 tok/s 2,520–6,720 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	1.0 GB	Q8_0	Comfortable
3,500 tok/s 2,100–5,600 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	1.0 GB	Q8_0	Comfortable
3,425 tok/s 2,055–5,480 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	1.0 GB	Q8_0	Comfortable
3,349 tok/s 2,009–5,358 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	1.0 GB	Q8_0	Comfortable
3,349 tok/s 2,009–5,358 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	1.0 GB	Q8_0	Comfortable
3,349 tok/s 2,009–5,358 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	1.0 GB	Q8_0	Comfortable
3,349 tok/s 2,009–5,358 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	1.0 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: IBM
Organisation type: Industry
Country: United States of America
Published: 17 September 2025

What it does

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

Domain: Vision, Multimodal, Language
Task: Visual question answering, Character recognition (OCR), Retrieval-augmented generation
Base model: SigLIP 2

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: 258M
Training data: tokens

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 H100 SXM5 80GB
Cloud vendor: IBM
Data centre: "We train granite-docling-258m using IBM's super computing cluster, Blue Vela, which is outfitted with NVIDIA H100 GPUs. This cluster provides a scalable and efficient infrastructure for training our models over thousands of GPUs."

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: Unreleased
Hugging Face: ibm-granite

How it is classified

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

Record confidence: Confident

Sources

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

Reference: IBM Granite-Docling: End-to-end document understanding with one tiny model
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run Granite-Docling

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 Granite-Docling

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

Tesla C1080

Memory needed

1.0 GB

Fastest

13,133 tok/s

Granite-Docling is small enough at 258M parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The smallest card that holds it is the Tesla C1080 with 4 GB, running it at Q8_0 and producing around 143 tokens per second.

The quickest result comes from a B200 at around 13,133 tokens per second — its 8,000 GB/s of bandwidth is what buys that.

Background

Granite-Docling was published by IBM, in United States of America, in September 2025. The organisation is categorised as industry.

It works in Vision, Multimodal, Language, and is recorded as doing visual question answering, Character recognition (OCR), Retrieval-augmented generation.

It is derived from SigLIP 2 rather than trained from scratch, which is the usual way a specialised model is produced.

Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded. It is published under the ibm-granite organisation on Hugging Face.

Reading the throughput figures

The median result is around 368.8 tokens per second; 818 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.

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 Granite-Docling

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

Look at what Granite-Docling actually needs — around 1.0 GB at Q8_0. No amount of processing power compensates for a card that cannot hold it.
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 Granite-Docling.
03

Choose how far you will compress it

Each card runs the least-compressed copy it can hold — Q8_0 on the smallest card that fits. Setting a floor drops the cards that only manage Granite-Docling by squeezing it further than you would want.
04

Sort by speed

The speed ordering for Granite-Docling is effectively an ordering by memory bandwidth, which is why the B200 tops it at 13,133 tok/s.
05

Look at the headroom, not just the fit

The fit column separates cards that just manage Granite-Docling 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 Granite-Docling.

Answers

Granite-Docling — common questions

Can I run Granite-Docling if it does not fit in my GPU?

It can be split between the card and system memory, but Granite-Docling generates painfully slowly that way. Nothing on this page assumes offloading.

Would two GPUs run Granite-Docling faster?

A second card roughly doubles the memory available but not the generation rate. With 818 cards already able to run Granite-Docling alone, the case for pairing is weak.

Why does the quantisation differ between cards for Granite-Docling?

Because capacity varies, so does how hard Granite-Docling has to be squeezed — 1 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these Granite-Docling speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 7,880–21,012 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.

What GPU do I need to run Granite-Docling?

The smallest card in our catalogue that holds Granite-Docling is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 1.0 GB, and produces roughly 143 tokens per second. 818 cards in total can run it.

How fast is Granite-Docling on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 13,133 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 818 of the cards that can run Granite-Docling clear that.

How much VRAM does Granite-Docling need?

About 1.0 GB at Q8_0 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 Granite-Docling on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 1.0 GB and generating roughly 2,446 tokens per second — a comfortable fit.

Can I run Granite-Docling on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 1.0 GB and generating roughly 1,498 tokens per second — a comfortable fit.

Can I run Granite-Docling on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 1.0 GB and generating roughly 1,855 tokens per second — a comfortable fit.

Can I run Granite-Docling on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 1.0 GB and generating roughly 2,200 tokens per second — a comfortable fit.

Is Granite-Docling open source?

Its weights are published, so Granite-Docling 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 Granite-Docling have?

Granite-Docling has 258M parameters. 258M. 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 Granite-Docling?

Granite-Docling was published by IBM, based in United States of America, categorised as industry.

When was Granite-Docling released?

Granite-Docling was published in September 2025.

What is Granite-Docling used for?

Granite-Docling works in Vision, Multimodal, Language, and is recorded as handling visual question answering, Character recognition (OCR), Retrieval-augmented generation. 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.

Where can I download Granite-Docling?

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

Source

Original publication

Record last updated 28 November 2025

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