GPT-J-6B TPS calculator

Open weights EleutherAI,LAION 6.1B parameters May 2021

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

589 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla K20c

5 GB · Q4_K_M · 28.6 tok/s

Fastest card

B200

560 tok/s · 180 GB

Which GPUs can run GPT-J-6B?

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

589 cards match

Calculating

					Needs	Quantisation	Fit
560 tok/s 336–896 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	7.2 GB	Q8_0	Comfortable
560 tok/s 336–896 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	7.2 GB	Q8_0	Comfortable
447 tok/s 268–715 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	7.2 GB	Q8_0	Comfortable
447 tok/s 268–715 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	7.2 GB	Q8_0	Comfortable
357 tok/s 214–572 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	7.2 GB	Q8_0	Comfortable
342 tok/s 205–547 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	7.2 GB	Q8_0	Comfortable
342 tok/s 205–547 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	7.2 GB	Q8_0	Comfortable
327 tok/s 196–524 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	7.2 GB	Q8_0	Comfortable
291 tok/s 174–465 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	7.2 GB	Q8_0	Comfortable
291 tok/s 174–465 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	7.2 GB	Q8_0	Comfortable
291 tok/s 174–465 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	7.2 GB	Q8_0	Comfortable
276 tok/s 165–441 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	7.2 GB	Q8_0	Comfortable
235 tok/s 141–376 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	7.2 GB	Q8_0	Comfortable
235 tok/s 141–376 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	7.2 GB	Q8_0	Comfortable
235 tok/s 141–376 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	7.2 GB	Q8_0	Comfortable
235 tok/s 141–376 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	7.2 GB	Q8_0	Comfortable
235 tok/s 141–376 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	7.2 GB	Q8_0	Comfortable
179 tok/s 107–286 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	7.2 GB	Q8_0	Comfortable
179 tok/s 107–286 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	7.2 GB	Q8_0	Comfortable
149 tok/s 90–239 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	7.2 GB	Q8_0	Comfortable
146 tok/s 88–234 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	7.2 GB	Q8_0	Comfortable
143 tok/s 86–228 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	7.2 GB	Q8_0	Comfortable
143 tok/s 86–228 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	7.2 GB	Q8_0	Comfortable
143 tok/s 86–228 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	7.2 GB	Q8_0	Comfortable
143 tok/s 86–228 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	7.2 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: EleutherAI,LAION
Organisation type: Research collective,Research collective
Country: United States of America, Germany
Published: 1 May 2021
Authors: Ben Wang, Aran Komatsuzaki

What it does

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

Domain: Language
Task: Language modeling/generation, Question answering, Automated theorem proving, Code generation
Approach: Self-supervised learning

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: 6.1B
Training data: 400,000,000,000 tokens
Epochs: 1

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: 1.5 × 10²² FLOP
How it was established: Reported

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: Google TPU v3
Wall-clock time: 840 hours (35 days)
Hardware utilisation: HFU 60.0%

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

How it is classified

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

Record confidence: Confident
Benchmark data: GPT-J-6B

Sources

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

Reference: GPT-J-6B: 6B JAX-Based Transformer
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run GPT-J-6B

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 GPT-J-6B

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

The hardware side

Minimum card

Tesla K20c

Memory needed

4.4 GB

Fastest

560 tok/s

GPT-J-6B is small enough at 6.1B parameters that hardware is rarely the obstacle — 589 of the cards we track can run it, including cards several years old.

At the low end, a Tesla K20c handles it — 5 GB, at Q4_K_M, for about 28.6 tokens per second.

A B200 is the fastest we calculate for it: about 560 tokens per second, from 8,000 GB/s of memory bandwidth.

What this model is

GPT-J-6B was published by EleutherAI,LAION, in United States of America, in May 2021. The organisation is categorised as research collective,Research collective.

It works in Language, and is recorded as doing language modeling/generation, Question answering, Automated theorem proving, Code generation.

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

What decides the speed

Across every card that can run it, the middle of the range is about 25.2 tokens per second, and 539 of them clear the ten tokens per second that roughly matches reading speed.

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

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.

What went into building it

Training it took roughly 1.5 × 10²² FLOP of computation, on Google TPU v3 — a measure of what producing the model cost, not of how fast it answers.

It was trained on about 400,000,000,000 tokens of text.

Step by step

How to choose a GPU for GPT-J-6B

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

01

Read the memory figure first

The table lists every card that can hold GPT-J-6B — around 4.4 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 GPT-J-6B.
03

Choose how far you will compress it

The quantisation column varies by card, because a bigger card holds a more accurate copy of GPT-J-6B — Q4_K_M on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Compare tokens per second, not specifications

Sort by speed to see how cards rank for GPT-J-6B. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 560 tok/s.
05

Check the fit verdict before buying

The fit column separates cards that just manage GPT-J-6B from those with room to spare. Buy for the second if the context might grow.
06

Check the card from the other side

Every card name links to its own page, which runs the same calculation across the whole model catalogue. Worth a look before buying for GPT-J-6B alone — a card is usually bought for more than one model.

Answers

GPT-J-6B — common questions

What GPU do I need to run GPT-J-6B?

The smallest card in our catalogue that holds GPT-J-6B is the Tesla K20c, with 5 GB of memory. It runs the model at Q4_K_M using about 4.4 GB, and produces roughly 28.6 tokens per second. 589 cards in total can run it.

How fast is GPT-J-6B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 560 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 539 of the cards that can run GPT-J-6B clear that.

How much VRAM does GPT-J-6B need?

About 4.4 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 GPT-J-6B on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 7.2 GB and generating roughly 104 tokens per second — a tight fit.

Can I run GPT-J-6B on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 7.2 GB and generating roughly 63.8 tokens per second — a comfortable fit.

Can I run GPT-J-6B on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 7.2 GB and generating roughly 79.1 tokens per second — a comfortable fit.

Can I run GPT-J-6B on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 7.2 GB and generating roughly 93.8 tokens per second — a comfortable fit.

Is GPT-J-6B open source?

Its weights are published, so GPT-J-6B 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 GPT-J-6B have?

GPT-J-6B has 6.1B parameters. source: model details table in GitHub. 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 GPT-J-6B?

GPT-J-6B was published by EleutherAI,LAION, based in United States of America, categorised as research collective,Research collective.

When was GPT-J-6B released?

GPT-J-6B was published in May 2021. 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 GPT-J-6B used for?

GPT-J-6B works in Language, and is recorded as handling language modeling/generation, Question answering, Automated theorem proving, Code generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download GPT-J-6B?

The weights for GPT-J-6B are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

How much compute was used to train GPT-J-6B?

Around 1.5 × 10²² FLOP, on Google TPU v3. 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 GPT-J-6B 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 0.8 GB. Our figures for GPT-J-6B assume it is fully resident.

Would two GPUs run GPT-J-6B faster?

A second card roughly doubles the memory available but not the generation rate. With 589 cards already able to run GPT-J-6B alone, the case for pairing is weak.

Why does the quantisation differ between cards for GPT-J-6B?

A larger card holds a more accurate copy. Across the cards that run GPT-J-6B, 3 compression levels are used; the floor control above pins it to one.

How accurate are these GPT-J-6B speed estimates?

These are estimates with real error bars. The fastest result here, 336–896 tok/s on the B200, could reasonably land anywhere in its published range depending on which runtime you use.

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