Pixtral Large TPS calculator
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
Smallest card that fits
RTX PRO 5000 72 GB Blackwell
72 GB · Q3_K_M · 12.4 tok/s
Fastest card
H100 NVL 94 GB
31.1 tok/s · 94 GB
Which GPUs can run Pixtral Large?
Set the inputs, read the answer
A longer conversation needs more memory, which can push this model off smaller cards.
Hides cards that would only fit the model by compressing it below this point.
38 cards match
Calculating| Needs | Quantisation | Fit | |||||
|---|---|---|---|---|---|---|---|
|
31.1
tok/s
19–50 · low confidence |
H100 NVL 94 GB NVIDIA | 94 GB | 3,940 GB/s | Mar 2023 | 75.7 GB | Q4_K_M | Tight |
|
28.2
tok/s
17–45 · low confidence |
H100 SXM5 80 GB NVIDIA | 80 GB | 3,360 GB/s | Oct 2022 | 68.5 GB | IQ4_XS | Tight |
|
28.2
tok/s
17–45 · low confidence |
H800 SXM5 NVIDIA | 80 GB | 3,360 GB/s | Mar 2023 | 68.5 GB | IQ4_XS | Tight |
|
27.3
tok/s
16–44 · low confidence |
B200 NVIDIA | 180 GB | 8,000 GB/s | Jan 2024 | 133.5 GB | Q8_0 | Comfortable |
|
27.3
tok/s
16–44 · low confidence |
B300 NVIDIA | 288 GB | 8,000 GB/s | Sep 2025 | 133.5 GB | Q8_0 | Comfortable |
|
26.5
tok/s
16–42 · low confidence |
H100 PCIe 96 GB NVIDIA | 96 GB | 3,360 GB/s | Mar 2023 | 75.7 GB | Q4_K_M | Tight |
|
26.5
tok/s
16–42 · low confidence |
H100 SXM5 94 GB NVIDIA | 94 GB | 3,360 GB/s | Mar 2023 | 75.7 GB | Q4_K_M | Tight |
|
26.5
tok/s
16–42 · low confidence |
H100 SXM5 96 GB NVIDIA | 96 GB | 3,360 GB/s | Mar 2023 | 75.7 GB | Q4_K_M | Tight |
|
25.4
tok/s
15–41 · low confidence |
Radeon Instinct MI300 AMD | 128 GB | 6,550 GB/s | Jan 2023 | 104.6 GB | Q6_K | Tight |
|
24.3
tok/s
15–39 · low confidence |
H200 NVL NVIDIA | 141 GB | 4,890 GB/s | Nov 2024 | 104.6 GB | Q6_K | Comfortable |
|
24.3
tok/s
15–39 · low confidence |
H200 SXM 141 GB NVIDIA | 141 GB | 4,890 GB/s | Nov 2024 | 104.6 GB | Q6_K | Comfortable |
|
21.8
tok/s
13–35 · low confidence |
Radeon Instinct MI350X AMD | 288 GB | 8,190 GB/s | Jan 2025 | 133.5 GB | Q8_0 | Comfortable |
|
21.8
tok/s
13–35 · low confidence |
Radeon Instinct MI355X AMD | 288 GB | 8,190 GB/s | Jan 2025 | 133.5 GB | Q8_0 | Comfortable |
|
20.6
tok/s
12–33 · low confidence |
Radeon Instinct MI300A AMD | 128 GB | 5,325 GB/s | Dec 2023 | 104.6 GB | Q6_K | Tight |
|
17.1
tok/s
10–27 · low confidence |
A100 SXM4 80 GB NVIDIA | 80 GB | 2,040 GB/s | Nov 2020 | 68.5 GB | IQ4_XS | Tight |
|
17.1
tok/s
10–27 · low confidence |
A100X NVIDIA | 80 GB | 2,040 GB/s | Jun 2021 | 68.5 GB | IQ4_XS | Tight |
|
17.1
tok/s
10–27 · low confidence |
A800 SXM4 80 GB NVIDIA | 80 GB | 2,040 GB/s | Aug 2022 | 68.5 GB | IQ4_XS | Tight |
|
17.1
tok/s
10–27 · low confidence |
H100 CNX NVIDIA | 80 GB | 2,040 GB/s | Mar 2023 | 68.5 GB | IQ4_XS | Tight |
|
17.1
tok/s
10–27 · low confidence |
H100 PCIe 80 GB NVIDIA | 80 GB | 2,040 GB/s | Oct 2022 | 68.5 GB | IQ4_XS | Tight |
|
17.1
tok/s
10–27 · low confidence |
H800 PCIe 80 GB NVIDIA | 80 GB | 2,040 GB/s | Mar 2023 | 68.5 GB | IQ4_XS | Tight |
|
16.3
tok/s
10–26 · low confidence |
A100 PCIe 80 GB NVIDIA | 80 GB | 1,940 GB/s | Jun 2021 | 68.5 GB | IQ4_XS | Tight |
|
16.3
tok/s
10–26 · low confidence |
A800 PCIe 80 GB NVIDIA | 80 GB | 1,940 GB/s | Nov 2022 | 68.5 GB | IQ4_XS | Tight |
|
16.0
tok/s
10–26 · low confidence |
Radeon Instinct MI325X AMD | 256 GB | 6,000 GB/s | Oct 2024 | 133.5 GB | Q8_0 | Comfortable |
|
14.2
tok/s
9–23 · low confidence |
Radeon Instinct MI300X AMD | 192 GB | 5,325 GB/s | Dec 2023 | 133.5 GB | Q8_0 | Comfortable |
|
14.2
tok/s
9–23 · low confidence |
Radeon Instinct MI308X AMD | 192 GB | 5,325 GB/s | Dec 2023 | 133.5 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
- Mistral AI
- Organisation type
- Industry
- Country
- France
- Published
- 18 November 2024
- Authors
- Albert Jiang, Alexandre Sablayrolles, Alexis Tacnet, Alok Kothari, Antoine Roux, Arthur Mensch, Audrey Herblin-Stoop, Augustin Garreau, Austin Birky, Bam4d, Baptiste Bout, Baudouin de Monicault, Blanche Savary, Carole Rambaud, Caroline Feldman, Devendra Singh Chaplot, Diego de las Casas, Diogo Costa, Eleonore Arcelin, Emma Bou Hanna, Etienne Metzger, Gaspard Blanchet, Gianna Lengyel, Guillaume Bou…
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Multimodal, Language, Vision
- Task
- Vision-language generation, Visual question answering, Mathematical reasoning, Character recognition (OCR), Language modeling/generation, Question answering
- Approach
- Supervised
- Base model
- Mistral Large 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
- 124B
- Training data
- tokens
123B multimodal decoder, 1B parameter vision encoder
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
mrl license (research only), separate license is needed for commercial usage https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Why it is tracked
- Significant use,SOTA improvement
- Record confidence
- Confident
Number of downloads not visible "State-of-the-art on MathVista, DocVQA, VQAv2"
Sources
Where this record came from and when it was last checked.
- Reference
- Pixtral Large
- Last updated
- 28 November 2025
The extremes
The ten fastest GPUs for Pixtral Large
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.
- 01 H100 NVL 94 GB 94 GB · 3,940 GB/s · Q4_K_M 31.1 tok/s
- 02 H800 SXM5 80 GB · 3,360 GB/s · IQ4_XS 28.2 tok/s
- 03 H100 SXM5 80 GB 80 GB · 3,360 GB/s · IQ4_XS 28.2 tok/s
- 04 B300 288 GB · 8,000 GB/s · Q8_0 27.3 tok/s
- 05 B200 180 GB · 8,000 GB/s · Q8_0 27.3 tok/s
- 06 H100 PCIe 96 GB 96 GB · 3,360 GB/s · Q4_K_M 26.5 tok/s
- 07 H100 SXM5 94 GB 94 GB · 3,360 GB/s · Q4_K_M 26.5 tok/s
- 08 H100 SXM5 96 GB 96 GB · 3,360 GB/s · Q4_K_M 26.5 tok/s
- 09 Radeon Instinct MI300 128 GB · 6,550 GB/s · Q6_K 25.4 tok/s
- 10 H200 NVL 141 GB · 4,890 GB/s · Q6_K 24.3 tok/s
The smallest GPUs that still run Pixtral Large
The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.
- 01 RTX PRO 5000 72 GB Blackwell 72 GB · needs 61.3 GB · Q3_K_M · tight 12.4 tok/s
- 02 H100 CNX 80 GB · needs 68.5 GB · IQ4_XS · tight 17.1 tok/s
- 03 H800 PCIe 80 GB 80 GB · needs 68.5 GB · IQ4_XS · tight 17.1 tok/s
- 04 H800 SXM5 80 GB · needs 68.5 GB · IQ4_XS · tight 28.2 tok/s
- 05 A800 PCIe 80 GB 80 GB · needs 68.5 GB · IQ4_XS · tight 16.3 tok/s
- 06 H100 PCIe 80 GB 80 GB · needs 68.5 GB · IQ4_XS · tight 17.1 tok/s
- 07 H100 SXM5 80 GB 80 GB · needs 68.5 GB · IQ4_XS · tight 28.2 tok/s
- 08 A800 SXM4 80 GB 80 GB · needs 68.5 GB · IQ4_XS · tight 17.1 tok/s
- 09 A100 PCIe 80 GB 80 GB · needs 68.5 GB · IQ4_XS · tight 16.3 tok/s
- 10 A100X 80 GB · needs 68.5 GB · IQ4_XS · tight 17.1 tok/s
What the numbers mean
The hardware side
Minimum card
RTX PRO 5000 72 GB Blackwell
Memory needed
61.3 GB
Fastest
31.1 tok/s
Pixtral Large sits at 124B parameters, which puts it above consumer hardware and into the range where a card is bought for this purpose rather than repurposed for it. 38 of the cards we track can hold it.
At the low end, a RTX PRO 5000 72 GB Blackwell handles it — 72 GB, at Q3_K_M, for about 12.4 tokens per second.
At the other end, a H100 NVL 94 GB generates roughly 31.1 tokens per second on it, on the strength of 3,940 GB/s of memory bandwidth.
About this model
Pixtral Large was published by Mistral AI, in France, in November 2024. It comes out of industry.
It works in Multimodal, Language, Vision, and is recorded as doing vision-language generation, Visual question answering, Mathematical reasoning, Character recognition (OCR), Language modeling/generation, Question answering.
It is derived from Mistral Large 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 mistralai organisation on Hugging Face.
How fast it runs, and why
The median result is around 17.1 tokens per second; 36 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.
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.
What went into building it
It is tracked in the underlying dataset for one reason in particular: significant use,SOTA improvement.
Step by step
How to choose a GPU for Pixtral Large
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 Pixtral Large — around 61.3 GB at Q3_K_M. Capacity is the gate — a card either holds it or it does not.
-
02
Match the context to your actual use
The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect: at long context Pixtral Large can slip off a card that handles short questions easily.
-
03
Decide how much compression you will accept
Each card runs the least-compressed copy it can hold — Q3_K_M on the smallest card that fits. Setting a floor drops the cards that only manage Pixtral Large by squeezing it further than you would want.
-
04
Compare tokens per second, not specifications
Sort by speed to see how cards rank for Pixtral Large. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the H100 NVL 94 GB tops it at 31.1 tok/s.
-
05
Look at the headroom, not just the fit
A tight fit runs Pixtral Large but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
-
06
Check the card from the other side
Following a card through to its own page shows every other model it can hold, which is the question that follows once Pixtral Large is settled.
Answers
Pixtral Large — common questions
Who created Pixtral Large?
Pixtral Large was published by Mistral AI, based in France, categorised as industry.
When was Pixtral Large released?
Pixtral Large was published in November 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 Pixtral Large used for?
Pixtral Large works in Multimodal, Language, Vision, and is recorded as handling vision-language generation, Visual question answering, Mathematical reasoning, Character recognition (OCR), Language modeling/generation, Question answering. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
Where can I download Pixtral Large?
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 Pixtral Large if it does not fit in my GPU?
Partly. Layers that do not fit sit in system memory and run at a fraction of the speed, so a mostly-offloaded Pixtral Large is rarely worth using — the nearest miss we calculate is short by 18.1 GB. Every figure here assumes the whole model is on the card.
Would two GPUs run Pixtral Large faster?
Two cards buy memory rather than speed. That matters for Pixtral Large only if one card cannot hold it — 38 can, so a second adds little.
Why does the quantisation differ between cards for Pixtral Large?
A larger card holds a more accurate copy. Across the cards that run Pixtral Large, 5 compression levels are used; the floor control above pins it to one.
How accurate are these Pixtral Large speed estimates?
Every figure is derived from memory bandwidth and model size, not benchmarked. That is why each is published as a range such as 19–50 tok/s on the H100 NVL 94 GB rather than a single number.
What GPU do I need to run Pixtral Large?
The smallest card in our catalogue that holds Pixtral Large is the RTX PRO 5000 72 GB Blackwell, with 72 GB of memory. It runs the model at Q3_K_M using about 61.3 GB, and produces roughly 12.4 tokens per second. 38 cards in total can run it.
How fast is Pixtral Large on a GPU?
It depends on the card. The quickest we calculate is a H100 NVL 94 GB at about 31.1 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 36 of the cards that can run Pixtral Large clear that.
How much VRAM does Pixtral Large need?
About 61.3 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.
Is Pixtral Large open source?
Its weights are published, so Pixtral Large 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 Pixtral Large have?
Pixtral Large has 124B parameters. 123B multimodal decoder, 1B parameter vision encoder. 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.
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.