GigaChat Lite (GigaChat-20B-A3B) TPS calculator

Open weights Sber 20B parameters December 2024

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

295 cards that can run it

818 cards we hold specifications for

Smallest card that fits

GeForce GTX 1080 Ti

11 GB · Q3_K_M · 131 tok/s

Fastest card

B200

941 tok/s · 180 GB

Which GPUs can run GigaChat Lite (GigaChat-20B-A3B)?

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

295 cards match

Calculating

					Needs	Quantisation	Fit
941 tok/s 565–1,506 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	21.3 GB	Q8_0	Comfortable
941 tok/s 565–1,506 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	21.3 GB	Q8_0	Comfortable
752 tok/s 451–1,202 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	21.3 GB	Q8_0	Comfortable
752 tok/s 451–1,202 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	21.3 GB	Q8_0	Comfortable
601 tok/s 361–962 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	21.3 GB	Q8_0	Comfortable
575 tok/s 345–920 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	21.3 GB	Q8_0	Comfortable
575 tok/s 345–920 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	21.3 GB	Q8_0	Comfortable
551 tok/s 330–881 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	21.3 GB	Q8_0	Comfortable
489 tok/s 293–782 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	21.3 GB	Q8_0	Comfortable
489 tok/s 293–782 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	21.3 GB	Q8_0	Comfortable
489 tok/s 293–782 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	21.3 GB	Q8_0	Comfortable
464 tok/s 278–742 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	21.3 GB	Q8_0	Comfortable
395 tok/s 237–632 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	21.3 GB	Q8_0	Comfortable
395 tok/s 237–632 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	21.3 GB	Q8_0	Comfortable
395 tok/s 237–632 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	21.3 GB	Q8_0	Comfortable
395 tok/s 237–632 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	21.3 GB	Q8_0	Comfortable
395 tok/s 237–632 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	21.3 GB	Q8_0	Comfortable
301 tok/s 181–482 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	21.3 GB	Q8_0	Comfortable
301 tok/s 181–482 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	21.3 GB	Q8_0	Comfortable
290 tok/s 174–463 · low confidence	GeForce RTX 3080 12 GB NVIDIA	12 GB	912 GB/s	Jan 2022	9.7 GB	Q3_K_M	Tight
290 tok/s 174–463 · low confidence	GeForce RTX 3080 Ti NVIDIA	12 GB	912 GB/s	May 2021	9.7 GB	Q3_K_M	Tight
251 tok/s 150–401 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	21.3 GB	Q8_0	Comfortable
245 tok/s 147–393 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	21.3 GB	Q8_0	Comfortable
244 tok/s 146–390 · low confidence	RTX A5000-12Q NVIDIA	12 GB	768 GB/s	Apr 2021	9.7 GB	Q3_K_M	Tight
240 tok/s 144–384 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	21.3 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: Sber
Organisation type: Industry,Government
Country: Russia
Published: 13 December 2024

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, Chat

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: 20B
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.

Training compute: 9.9 × 10²² FLOP

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: ai-sage

How it is classified

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

Record confidence: Speculative

Sources

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

Reference: the first open MoE model in Russia
Last updated: 11 February 2026

The extremes

The ten fastest GPUs that run GigaChat Lite (GigaChat-20B-A3B)

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 GigaChat Lite (GigaChat-20B-A3B)

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

GeForce GTX 1080 Ti

Memory needed

9.7 GB

Fastest

941 tok/s

GigaChat Lite (GigaChat-20B-A3B) reaches a parameter count of 20B. That lands in the range a serious desktop card can handle once the weights are compressed. The number of cards we track that can run it: 295.

The smallest card that holds it is GeForce GTX 1080 Ti, with a memory capacity of 11 GB, running it at a compression of Q3_K_M and producing around 131 tokens per second.

The fastest we calculate for it is B200, generating roughly 941 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

What this model is

GigaChat Lite (GigaChat-20B-A3B) was published by Sber, in the country recorded as Russia, during December 2024. The publishing organisation is categorised as industry,Government.

It works in the domain of Language, and is recorded as performing the task of language modeling/generation, Question answering, Chat.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached. On Hugging Face it is published under the organisation ai-sage.

What decides the speed

Across every card that can run it, the middle of the range sits at 101.7 tokens per second. Exceeding reading speed outright: 293 of them.

This is a mixture-of-experts model, which routes each token through only part of itself. It therefore generates far faster than its total size suggests — while still needing every parameter resident in memory, so it is quick without being cheap to hold.

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 9.9 × 10²² FLOP. 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 GigaChat Lite (GigaChat-20B-A3B)

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 GigaChat Lite (GigaChat-20B-A3B), needing around 9.7 GB at a compression of Q3_K_M. No amount of processing power compensates for a card that cannot hold it.
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 a card that seemed fine stops fitting GigaChat Lite (GigaChat-20B-A3B).
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

Ranking by tokens per second follows memory bandwidth rather than core counts, for GigaChat Lite (GigaChat-20B-A3B). It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 941 tok/s.
05

Read the fit column last

A tight fit runs, but leaves nothing spare for a longer conversation, in the case of GigaChat Lite (GigaChat-20B-A3B). 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

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. A card is usually bought for more than one model, so it is worth a look before buying for GigaChat Lite (GigaChat-20B-A3B).

Answers

GigaChat Lite (GigaChat-20B-A3B) — common questions

GigaChat Lite (GigaChat-20B-A3B)— how much VRAM does it need?

It needs about 9.7 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.

GigaChat Lite (GigaChat-20B-A3B)— 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 Q3_K_M, using about 9.7 GB and generating roughly 290 tokens per second. The fit is tight.

GigaChat Lite (GigaChat-20B-A3B)— 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 Q5_K_M, using about 14.3 GB and generating roughly 237 tokens per second. The fit is tight.

GigaChat Lite (GigaChat-20B-A3B)— 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 21.3 GB and generating roughly 158 tokens per second. The fit is tight.

GigaChat Lite (GigaChat-20B-A3B)— 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.

GigaChat Lite (GigaChat-20B-A3B)— how many parameters does it have?

It has a parameter count of 20B. 20B total parameters 3.3B active 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.

GigaChat Lite (GigaChat-20B-A3B)— who created it?

It was published by Sber, based in Russia, an organisation categorised as industry,Government.

GigaChat Lite (GigaChat-20B-A3B)— when was it released?

It was published in December 2024. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

GigaChat Lite (GigaChat-20B-A3B)— what is it used for?

It works in the domain of Language, and is recorded as handling the task of language modeling/generation, Question answering, Chat. These are the areas it was designed around; they describe intent rather than a hard boundary.

GigaChat Lite (GigaChat-20B-A3B)— where can I download it?

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

GigaChat Lite (GigaChat-20B-A3B)— how much compute was used to train it?

Training consumed around 9.9 × 10²² FLOP. 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.

GigaChat Lite (GigaChat-20B-A3B)— 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 3.0 GB. Every figure here assumes the whole model is resident on the card.

GigaChat Lite (GigaChat-20B-A3B)— would two GPUs run it faster?

A second card roughly doubles the memory available but not the generation rate. The number already able to run it alone: 295. So a second card is rarely the answer here.

GigaChat Lite (GigaChat-20B-A3B)— 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.

GigaChat Lite (GigaChat-20B-A3B)— how accurate are these speed estimates?

They are calculated from specifications rather than measured, and each carries a range. One example: 565–1,506 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

GigaChat Lite (GigaChat-20B-A3B)— what GPU do I need to run it?

The smallest card in our catalogue that holds it is GeForce GTX 1080 Ti, with a memory capacity of 11 GB. It runs the model at a compression of Q3_K_M using about 9.7 GB, and produces roughly 131 tokens per second. The number of cards able to run it in total: 295.

GigaChat Lite (GigaChat-20B-A3B)— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 941 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: 293.

Source

Original publication

Record last updated 11 February 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