Codestral TPS calculator

Open weights Mistral AI 22.2B parameters May 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

241 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Xeon Phi 7120P

16 GB · Q4_K_M · 10.1 tok/s

Fastest card

B200

153 tok/s · 180 GB

Which GPUs can run Codestral?

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

241 cards match

Calculating

					Needs	Quantisation	Fit
153 tok/s 92–244 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	24.5 GB	Q8_0	Comfortable
153 tok/s 92–244 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	24.5 GB	Q8_0	Comfortable
122 tok/s 73–195 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	24.5 GB	Q8_0	Comfortable
122 tok/s 73–195 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	24.5 GB	Q8_0	Comfortable
97.5 tok/s 58–156 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	24.5 GB	Q8_0	Comfortable
93.3 tok/s 56–149 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	24.5 GB	Q8_0	Comfortable
93.3 tok/s 56–149 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	24.5 GB	Q8_0	Comfortable
89.3 tok/s 54–143 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	24.5 GB	Q8_0	Comfortable
79.2 tok/s 48–127 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	24.5 GB	Q8_0	Comfortable
79.2 tok/s 48–127 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	24.5 GB	Q8_0	Comfortable
79.2 tok/s 48–127 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	24.5 GB	Q8_0	Comfortable
75.2 tok/s 45–120 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	24.5 GB	Q8_0	Comfortable
64.1 tok/s 38–103 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	24.5 GB	Q8_0	Comfortable
64.1 tok/s 38–103 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	24.5 GB	Q8_0	Comfortable
64.1 tok/s 38–103 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	24.5 GB	Q8_0	Comfortable
64.1 tok/s 38–103 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	24.5 GB	Q8_0	Comfortable
64.1 tok/s 38–103 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	24.5 GB	Q8_0	Comfortable
49.8 tok/s 30–80 · low confidence	Tesla V100 SXM2 16 GB NVIDIA	16 GB	1,130 GB/s	Nov 2019	14.1 GB	Q4_K_M	Tight
48.8 tok/s 29–78 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	24.5 GB	Q8_0	Comfortable
48.8 tok/s 29–78 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	24.5 GB	Q8_0	Comfortable
42.3 tok/s 25–68 · low confidence	GeForce RTX 5080 NVIDIA	16 GB	960 GB/s	Jan 2025	14.1 GB	Q4_K_M	Tight
40.7 tok/s 24–65 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	24.5 GB	Q8_0	Comfortable
39.8 tok/s 24–64 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	24.5 GB	Q8_0	Comfortable
39.5 tok/s 24–63 · low confidence	Tesla V100 DGXS 16 GB NVIDIA	16 GB	897 GB/s	Mar 2018	14.1 GB	Q4_K_M	Tight
39.5 tok/s 24–63 · low confidence	Tesla V100 PCIe 16 GB NVIDIA	16 GB	897 GB/s	Jun 2017	14.1 GB	Q4_K_M	Tight

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: Mistral AI
Organisation type: Industry
Country: France
Published: 29 May 2024
Authors: Albert Jiang, Alexandre Sablayrolles, Alexis Tacnet, Antoine Roux, Arthur Mensch, Audrey Herblin-Stoop, Baptiste Bout, Baudouin de Monicault, Blanche Savary, Bam4d, Caroline Feldman, Devendra Singh Chaplot, Diego de las Casas, Eleonore Arcelin, Emma Bou Hanna, Etienne Metzger, Gianna Lengyel, Guillaume Bour, Guillaume Lample, Harizo Rajaona, Henri Roussez, Jean-Malo Delignon, Jia Li, Justus Murke,…

What it does

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

Domain: Language
Task: Code generation, Code autocompletion

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: 22.2B
Training data: tokens

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 (non-commercial)
Training code: Unreleased
Hugging Face: mistralai

How it is classified

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

Likely above 10²³ FLOP: Yes
Record confidence: Confident

Sources

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

Reference: Empowering developers and democratising coding with Mistral AI.
Last updated: 19 June 2026

The extremes

The ten fastest GPUs that run Codestral

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 Codestral

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 you need to run it

Minimum card

Xeon Phi 7120P

Memory needed

14.1 GB

Fastest

153 tok/s

With 22.2B parameters, Codestral lands in the range a serious desktop card can handle once the weights are compressed. 241 of the cards we track can run it.

The least hardware that works is a Xeon Phi 7120P. Its 16 GB is enough at Q4_K_M compression, giving roughly 10.1 tokens per second.

Top of the range is the B200, at roughly 153 tokens per second thanks to 8,000 GB/s of bandwidth.

Background

Codestral was published by Mistral AI, in France, in May 2024. The organisation is categorised as industry.

It works in Language, and is recorded as doing code generation, Code autocompletion.

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all. It is published under the mistralai organisation on Hugging Face.

Reading the throughput figures

The median result is around 19.7 tokens per second; 194 cards produce text faster than most people read it.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

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 Codestral

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 Codestral — around 14.1 GB at Q4_K_M. That figure, not the card's headline performance, is what decides whether it runs.
02

Decide how long your conversations run

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for Codestral.
03

Decide how much compression you will accept

Compression is what makes Codestral fit smaller cards, at some cost in accuracy — Q4_K_M on the smallest card that fits. A minimum quality removes the ones that go too far.
04

Sort by speed

Sort by speed to see how cards rank for Codestral. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 153 tok/s.
05

Read the fit column last

A tight fit runs Codestral but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
06

Open the card you have settled on

Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond Codestral.

Answers

Codestral — common questions

Would two GPUs run Codestral faster?

Capacity adds across cards; throughput does not. Since 241 of the cards we track already hold Codestral on their own, a second card is rarely the answer here.

Why does the quantisation differ between cards for Codestral?

Because capacity varies, so does how hard Codestral has to be squeezed — 4 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these Codestral speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 92–244 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 Codestral?

The smallest card in our catalogue that holds Codestral is the Xeon Phi 7120P, with 16 GB of memory. It runs the model at Q4_K_M using about 14.1 GB, and produces roughly 10.1 tokens per second. 241 cards in total can run it.

How fast is Codestral on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 153 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 194 of the cards that can run Codestral clear that.

How much VRAM does Codestral need?

About 14.1 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 Codestral on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q4_K_M, using about 14.1 GB and generating roughly 49.8 tokens per second — a tight fit.

Can I run Codestral on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q6_K, using about 19.3 GB and generating roughly 37.1 tokens per second — a tight fit.

Is Codestral open source?

Its weights are published, so Codestral 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 Codestral have?

Codestral has 22.2B parameters. 22.2B from hugging face model card. 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 Codestral?

Codestral was published by Mistral AI, based in France, categorised as industry.

When was Codestral released?

Codestral was published in May 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.

What is Codestral used for?

Codestral works in Language, and is recorded as handling code generation, Code autocompletion. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download Codestral?

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

Can I run Codestral if it does not fit in my GPU?

It can be split between the card and system memory, but Codestral generates painfully slowly that way — the nearest miss we calculate is short by 3.3 GB. Nothing on this page assumes offloading.

Source

Original publication

Record last updated 19 June 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