Polyglot-Ko-12.8B TPS calculator

Open weights EleutherAI 12.9B parameters June 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 · 18.4 tok/s

Fastest card

B200

263 tok/s · 180 GB

Which GPUs can run Polyglot-Ko-12.8B?

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
263 tok/s 158–420 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	14.5 GB	Q8_0	Comfortable
263 tok/s 158–420 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	14.5 GB	Q8_0	Comfortable
210 tok/s 126–336 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	14.5 GB	Q8_0	Comfortable
210 tok/s 126–336 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	14.5 GB	Q8_0	Comfortable
168 tok/s 101–268 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	14.5 GB	Q8_0	Comfortable
161 tok/s 96–257 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	14.5 GB	Q8_0	Comfortable
161 tok/s 96–257 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	14.5 GB	Q8_0	Comfortable
154 tok/s 92–246 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	14.5 GB	Q8_0	Comfortable
136 tok/s 82–218 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	14.5 GB	Q8_0	Comfortable
136 tok/s 82–218 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	14.5 GB	Q8_0	Comfortable
136 tok/s 82–218 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	14.5 GB	Q8_0	Comfortable
132 tok/s 79–211 · low confidence	CMP 170HX 8 GB NVIDIA	8 GB	1,490 GB/s	Sep 2021	7.0 GB	Q3_K_M	Tight
129 tok/s 78–207 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	14.5 GB	Q8_0	Comfortable
118 tok/s 71–189 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.5 GB	Q4_K_M	Tight
110 tok/s 66–177 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	14.5 GB	Q8_0	Comfortable
110 tok/s 66–177 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	14.5 GB	Q8_0	Comfortable
110 tok/s 66–177 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	14.5 GB	Q8_0	Comfortable
110 tok/s 66–177 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	14.5 GB	Q8_0	Comfortable
110 tok/s 66–177 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	14.5 GB	Q8_0	Comfortable
84.0 tok/s 50–134 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	14.5 GB	Q8_0	Comfortable
84.0 tok/s 50–134 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	14.5 GB	Q8_0	Comfortable
70.0 tok/s 42–112 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	14.5 GB	Q8_0	Comfortable
68.5 tok/s 41–110 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	14.5 GB	Q8_0	Comfortable
68.1 tok/s 41–109 · low confidence	RTX A5000-8Q NVIDIA	8 GB	768 GB/s	Apr 2021	7.0 GB	Q3_K_M	Tight
67.0 tok/s 40–107 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	14.5 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
Organisation type: Research collective
Country: United States of America
Published: 4 June 2023
Authors: Hyunwoong Ko, Kichang Yang, Minho Ryu, Taekyoon Choi, Seungmu Yang, Jiwung Hyun, Sungho Park, Kyubyong Park

What it does

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

Domain: Language
Task: Translation, Language modeling/generation

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: 12.9B
Training data: 95,793,000,000 tokens
Batch size: 554,817

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: 1.3 × 10²² FLOP
How it was established: Operation counting

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
Chips used: 256
Power draw: 203.9 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 (unrestricted)

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: 35

Sources

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

Reference: A Technical Report for Polyglot-Ko: Open-Source Large-Scale Korean Language Models
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Polyglot-Ko-12.8B

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 Polyglot-Ko-12.8B

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

7.0 GB

Fastest

263 tok/s

Polyglot-Ko-12.8B is small enough at 12.9B parameters that hardware is rarely the obstacle — 509 of the cards we track can run it, including cards several years old.

The smallest card that holds it is the Xeon Phi 5110P with 8 GB, running it at Q3_K_M and producing around 18.4 tokens per second.

A B200 is the fastest we calculate for it: about 263 tokens per second, from 8,000 GB/s of memory bandwidth.

Where it came from

Polyglot-Ko-12.8B was published by EleutherAI, in United States of America, in June 2023. research collective is the category the publisher falls under.

It works in Language, and is recorded as doing translation, Language modeling/generation.

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 21.4 tokens per second; 459 cards produce text faster than most people read it.

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.

Training and provenance

Training it took roughly 1.3 × 10²² FLOP of computation, on NVIDIA A100 — a measure of what producing the model cost, not of how fast it answers.

It was trained on about 95,793,000,000 tokens of text.

Step by step

How to choose a GPU for Polyglot-Ko-12.8B

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

Every card here has been checked against Polyglot-Ko-12.8B — around 7.0 GB at Q3_K_M. Capacity is the gate — a card either holds it or it does not.
02

Set the context length you will work at

Set the context to what you will actually use. The cache grows with the conversation, and it is the usual reason Polyglot-Ko-12.8B stops fitting a card that seemed fine.
03

Decide how much compression you will accept

The quantisation column varies by card, because a bigger card holds a more accurate copy of Polyglot-Ko-12.8B — Q3_K_M on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Sort by speed

The speed ordering for Polyglot-Ko-12.8B is effectively an ordering by memory bandwidth, which is why the B200 tops it at 263 tok/s.
05

Look at the headroom, not just the fit

A tight fit runs Polyglot-Ko-12.8B but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
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 Polyglot-Ko-12.8B is settled.

Answers

Polyglot-Ko-12.8B — common questions

Where can I download Polyglot-Ko-12.8B?

The weights for Polyglot-Ko-12.8B 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 Polyglot-Ko-12.8B?

Around 1.3 × 10²² FLOP, on NVIDIA A100. 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 Polyglot-Ko-12.8B if it does not fit in my GPU?

It can be split between the card and system memory, but Polyglot-Ko-12.8B generates painfully slowly that way — the nearest miss we calculate is short by 3.1 GB. Nothing on this page assumes offloading.

Would two GPUs run Polyglot-Ko-12.8B faster?

Two cards buy memory rather than speed. That matters for Polyglot-Ko-12.8B only if one card cannot hold it — 509 can, so a second adds little.

Why does the quantisation differ between cards for Polyglot-Ko-12.8B?

Because capacity varies, so does how hard Polyglot-Ko-12.8B has to be squeezed — 5 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these Polyglot-Ko-12.8B speed estimates?

These are estimates with real error bars. The fastest result here, 158–420 tok/s on the B200, could reasonably land anywhere in its published range depending on which runtime you use.

What GPU do I need to run Polyglot-Ko-12.8B?

The smallest card in our catalogue that holds Polyglot-Ko-12.8B is the Xeon Phi 5110P, with 8 GB of memory. It runs the model at Q3_K_M using about 7.0 GB, and produces roughly 18.4 tokens per second. 509 cards in total can run it.

How fast is Polyglot-Ko-12.8B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 263 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 459 of the cards that can run Polyglot-Ko-12.8B clear that.

How much VRAM does Polyglot-Ko-12.8B need?

About 7.0 GB at Q3_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 Polyglot-Ko-12.8B on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q3_K_M, using about 7.0 GB and generating roughly 132 tokens per second — a tight fit.

Can I run Polyglot-Ko-12.8B on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q5_K_M, using about 10.0 GB and generating roughly 53.5 tokens per second — a tight fit.

Can I run Polyglot-Ko-12.8B on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q6_K, using about 11.5 GB and generating roughly 53.9 tokens per second — a comfortable fit.

Can I run Polyglot-Ko-12.8B on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 14.5 GB and generating roughly 44.0 tokens per second — a comfortable fit.

Is Polyglot-Ko-12.8B open source?

Its weights are published, so Polyglot-Ko-12.8B 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 Polyglot-Ko-12.8B have?

Polyglot-Ko-12.8B has 12.9B parameters. 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 Polyglot-Ko-12.8B?

Polyglot-Ko-12.8B was published by EleutherAI, based in United States of America, categorised as research collective.

When was Polyglot-Ko-12.8B released?

Polyglot-Ko-12.8B was published in June 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 Polyglot-Ko-12.8B used for?

Polyglot-Ko-12.8B works in Language, and is recorded as handling translation, Language modeling/generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

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