Chameleon-34B TPS calculator

Open weights Facebook AI Research 34B 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

132 cards that can run it

818 cards we hold specifications for

Smallest card that fits

RTX A4500

20 GB · Q3_K_M · 21.5 tok/s

Fastest card

B200

99.7 tok/s · 180 GB

Which GPUs can run Chameleon-34B?

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.

132 cards match

Calculating
Needs Quantisation Fit
99.7 tok/s

60–159 · low confidence

B200 NVIDIA 180 GB 8,000 GB/s Jan 2024 37.1 GB Q8_0 Comfortable
99.7 tok/s

60–159 · low confidence

B300 NVIDIA 288 GB 8,000 GB/s Sep 2025 37.1 GB Q8_0 Comfortable
79.6 tok/s

48–127 · low confidence

Radeon Instinct MI350X AMD 288 GB 8,190 GB/s Jan 2025 37.1 GB Q8_0 Comfortable
79.6 tok/s

48–127 · low confidence

Radeon Instinct MI355X AMD 288 GB 8,190 GB/s Jan 2025 37.1 GB Q8_0 Comfortable
63.6 tok/s

38–102 · low confidence

Radeon Instinct MI300 AMD 128 GB 6,550 GB/s Jan 2023 37.1 GB Q8_0 Comfortable
60.9 tok/s

37–97 · low confidence

H200 NVL NVIDIA 141 GB 4,890 GB/s Nov 2024 37.1 GB Q8_0 Comfortable
60.9 tok/s

37–97 · low confidence

H200 SXM 141 GB NVIDIA 141 GB 4,890 GB/s Nov 2024 37.1 GB Q8_0 Comfortable
58.3 tok/s

35–93 · low confidence

Radeon Instinct MI325X AMD 256 GB 6,000 GB/s Oct 2024 37.1 GB Q8_0 Comfortable
51.7 tok/s

31–83 · low confidence

Radeon Instinct MI300A AMD 128 GB 5,325 GB/s Dec 2023 37.1 GB Q8_0 Comfortable
51.7 tok/s

31–83 · low confidence

Radeon Instinct MI300X AMD 192 GB 5,325 GB/s Dec 2023 37.1 GB Q8_0 Comfortable
51.7 tok/s

31–83 · low confidence

Radeon Instinct MI308X AMD 192 GB 5,325 GB/s Dec 2023 37.1 GB Q8_0 Comfortable
49.1 tok/s

29–79 · low confidence

H100 NVL 94 GB NVIDIA 94 GB 3,940 GB/s Mar 2023 37.1 GB Q8_0 Comfortable
41.9 tok/s

25–67 · low confidence

H100 PCIe 96 GB NVIDIA 96 GB 3,360 GB/s Mar 2023 37.1 GB Q8_0 Comfortable
41.9 tok/s

25–67 · low confidence

H100 SXM5 80 GB NVIDIA 80 GB 3,360 GB/s Oct 2022 37.1 GB Q8_0 Comfortable
41.9 tok/s

25–67 · low confidence

H100 SXM5 94 GB NVIDIA 94 GB 3,360 GB/s Mar 2023 37.1 GB Q8_0 Comfortable
41.9 tok/s

25–67 · low confidence

H100 SXM5 96 GB NVIDIA 96 GB 3,360 GB/s Mar 2023 37.1 GB Q8_0 Comfortable
41.9 tok/s

25–67 · low confidence

H800 SXM5 NVIDIA 80 GB 3,360 GB/s Mar 2023 37.1 GB Q8_0 Comfortable
41.6 tok/s

25–67 · low confidence

DRIVE A100 PROD NVIDIA 32 GB 1,870 GB/s May 2020 25.2 GB Q5_K_M Tight
41.6 tok/s

25–67 · low confidence

GRID A100A NVIDIA 32 GB 1,870 GB/s May 2020 25.2 GB Q5_K_M Tight
39.8 tok/s

24–64 · low confidence

GeForce RTX 5090 NVIDIA 32 GB 1,790 GB/s Jan 2025 25.2 GB Q5_K_M Tight
39.8 tok/s

24–64 · low confidence

GeForce RTX 5090 D NVIDIA 32 GB 1,790 GB/s Jan 2025 25.2 GB Q5_K_M Tight
38.5 tok/s

23–62 · low confidence

GeForce RTX 5090 D V2 NVIDIA 24 GB 1,340 GB/s Aug 2025 21.3 GB Q4_K_M Tight
35.1 tok/s

21–56 · low confidence

A30X NVIDIA 24 GB 1,220 GB/s Apr 2021 21.3 GB Q4_K_M Tight
31.9 tok/s

19–51 · low confidence

Radeon Instinct MI250 AMD 128 GB 3,280 GB/s Nov 2021 37.1 GB Q8_0 Comfortable
31.9 tok/s

19–51 · low confidence

Radeon Instinct MI250X AMD 128 GB 3,280 GB/s Nov 2021 37.1 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
Facebook AI Research
Organisation type
Industry
Country
United States of America, France
Published
16 May 2024
Authors
Srinivasan Iyer, Bernie Huang, Lili Yu, Arun Babu, Chunting Zhou, Kushal Tirumala, Xi Victoria Lin, Hu Xu, Xian Li, Akshat Shrivastava, Omer Levy, Armen Aghajanyan, Ram Pasunuru, Andrew Cohen, Aram H. Markosyan, Koustuv Sinha, Xiaoqing Ellen Tan, Ivan Evtimov, Ping Yu, Tianlu Wang, Olga Golovneva, Asli Celikyilmaz, Pedro Rodriguez, Leonid Shamis, Vasu Sharma, Christine Jou, Karthik Padthe, Ching-F…

What it does

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

Domain
Multimodal, Image generation, Language, Vision
Task
Language modeling/generation, Vision-language generation, Visual question answering, Text-to-image
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
34B
Training data
4,400,000,000,000 tokens

Slightly conflicting info. Pre-training data details describe different types of data that sum to 4.8 trillion tokens, but Table 1 indicates 4.4T. Using table values as this agrees with other statements about epochs and total tokens seen.

Epochs
2.1

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
1.6 × 10²⁴ FLOP

GPU method: Table 2 shows that 34B model pre-training uses 4282407 GPU-hours, trained across 3072 A100s. 3.12e14 * 4282407 * 3600 * 0.3 = 1.44e24 Parameter-token method: Pre-training goes over 9.2T tokens, post-training only goes over 1.1B tokens (sum of tokens column in Table 3). 6 * 34B * 9.2T = 1.88e24 Geometric mean: sqrt(1.44e24 * 1.88e24) = 1.65e24

How it was established
Hardware,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 SXM4 80 GB
Chips used
3,072
Wall-clock time
1,394 hours (58.1 days)

34B model pre-training uses 4282407 GPU-hours, trained across 3072 A100s 4282407 / 3072 = 1394

Power draw
2.4 MW

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

https://ai.meta.com/resources/models-and-libraries/chameleon-downloads/?gk_enable=chameleon_web_flow_is_live "The models we’re releasing today were safety tuned and support mixed-modal inputs and text-only output to be used for research purposes. While we’ve taken steps to develop these models responsibly, we recognize that risks remain. At this time, we are not releasing the Chameleon image generation model."

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
Chameleon: Mixed-Modal Early-Fusion Foundation Models
Last updated
28 November 2025

The extremes

The ten fastest GPUs that run Chameleon-34B

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.

  1. 01 B300 288 GB · 8,000 GB/s · Q8_0 99.7 tok/s
  2. 02 B200 180 GB · 8,000 GB/s · Q8_0 99.7 tok/s
  3. 03 Radeon Instinct MI350X 288 GB · 8,190 GB/s · Q8_0 79.6 tok/s
  4. 04 Radeon Instinct MI355X 288 GB · 8,190 GB/s · Q8_0 79.6 tok/s
  5. 05 Radeon Instinct MI300 128 GB · 6,550 GB/s · Q8_0 63.6 tok/s
  6. 06 H200 NVL 141 GB · 4,890 GB/s · Q8_0 60.9 tok/s
  7. 07 H200 SXM 141 GB 141 GB · 4,890 GB/s · Q8_0 60.9 tok/s
  8. 08 Radeon Instinct MI325X 256 GB · 6,000 GB/s · Q8_0 58.3 tok/s
  9. 09 Radeon Instinct MI300A 128 GB · 5,325 GB/s · Q8_0 51.7 tok/s
  10. 10 Radeon Instinct MI300X 192 GB · 5,325 GB/s · Q8_0 51.7 tok/s

The smallest GPUs that still run Chameleon-34B

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

  1. 01 RTX 4000 Ada Generation 20 GB · needs 17.3 GB · Q3_K_M · tight 12.1 tok/s
  2. 02 RTX 4000 SFF Ada Generation 20 GB · needs 17.3 GB · Q3_K_M · tight 9.4 tok/s
  3. 03 Radeon RX 7900 XT 20 GB · needs 17.3 GB · Q3_K_M · tight 21.0 tok/s
  4. 04 A10M 20 GB · needs 17.3 GB · Q3_K_M · tight 16.8 tok/s
  5. 05 GeForce RTX 3080 Ti 20 GB 20 GB · needs 17.3 GB · Q3_K_M · tight 25.6 tok/s
  6. 06 RTX A4500 20 GB · needs 17.3 GB · Q3_K_M · tight 21.5 tok/s
  7. 07 Arc Pro B60 24 GB · needs 21.3 GB · Q4_K_M · tight 8.5 tok/s
  8. 08 GeForce RTX 5090 D V2 24 GB · needs 21.3 GB · Q4_K_M · tight 38.5 tok/s
  9. 09 RTX PRO 4000 Blackwell SFF 24 GB · needs 21.3 GB · Q4_K_M · tight 12.4 tok/s
  10. 10 GeForce RTX 5090 Mobile 24 GB · needs 21.3 GB · Q4_K_M · tight 25.8 tok/s

What the numbers mean

The hardware side

Minimum card

RTX A4500

Memory needed

17.3 GB

Fastest

99.7 tok/s

With 34B parameters, Chameleon-34B lands in the range a serious desktop card can handle once the weights are compressed. 132 of the cards we track can run it.

At the low end, a RTX A4500 handles it — 20 GB, at Q3_K_M, for about 21.5 tokens per second.

At the other end, a B200 generates roughly 99.7 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

Where it came from

Chameleon-34B was published by Facebook AI Research, in United States of America, in May 2024. The organisation is categorised as industry.

It works in Multimodal, Image generation, Language, Vision, and is recorded as doing language modeling/generation, Vision-language generation, Visual question answering, Text-to-image.

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

Understanding the speeds

Half the cards that hold it manage more than 21.2 tokens per second, and 103 exceed reading speed outright.

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.

Training and provenance

The training run consumed about 1.6 × 10²⁴ FLOP, on NVIDIA A100 SXM4 80 GB. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

The training set ran to roughly 4,400,000,000,000 tokens.

Step by step

How to choose a GPU for Chameleon-34B

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

  1. 01

    Start from the memory column

    Every card here has been checked against Chameleon-34B — around 17.3 GB at Q3_K_M. Capacity is the gate — a card either holds it or it does not.

  2. 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 Chameleon-34B.

  3. 03

    Choose how far you will compress it

    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 Chameleon-34B by squeezing it further than you would want.

  4. 04

    Sort by speed

    Sort by speed to see how cards rank for Chameleon-34B. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 99.7 tok/s.

  5. 05

    Look at the headroom, not just the fit

    A tight fit runs Chameleon-34B but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.

  6. 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 Chameleon-34B.

Answers

Chameleon-34B — common questions

01

How fast is Chameleon-34B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 99.7 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 103 of the cards that can run Chameleon-34B clear that.

02

How much VRAM does Chameleon-34B need?

About 17.3 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.

03

Can I run Chameleon-34B on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q4_K_M, using about 21.3 GB and generating roughly 38.5 tokens per second — a tight fit.

04

Is Chameleon-34B open source?

Its weights are published, so Chameleon-34B 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.

05

How many parameters does Chameleon-34B have?

Chameleon-34B has 34B 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.

06

Who created Chameleon-34B?

Chameleon-34B was published by Facebook AI Research, based in United States of America, categorised as industry.

07

When was Chameleon-34B released?

Chameleon-34B 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.

08

What is Chameleon-34B used for?

Chameleon-34B works in Multimodal, Image generation, Language, Vision, and is recorded as handling language modeling/generation, Vision-language generation, Visual question answering, Text-to-image. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

09

Where can I download Chameleon-34B?

The weights for Chameleon-34B are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

10

How much compute was used to train Chameleon-34B?

Around 1.6 × 10²⁴ FLOP, on NVIDIA A100 SXM4 80 GB. 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.

11

Can I run Chameleon-34B 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 6.9 GB. Our figures for Chameleon-34B assume it is fully resident.

12

Would two GPUs run Chameleon-34B faster?

A second card roughly doubles the memory available but not the generation rate. With 132 cards already able to run Chameleon-34B alone, the case for pairing is weak.

13

Why does the quantisation differ between cards for Chameleon-34B?

A larger card holds a more accurate copy. Across the cards that run Chameleon-34B, 5 compression levels are used; the floor control above pins it to one.

14

How accurate are these Chameleon-34B 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 60–159 tok/s on the B200 rather than a single number.

15

What GPU do I need to run Chameleon-34B?

The smallest card in our catalogue that holds Chameleon-34B is the RTX A4500, with 20 GB of memory. It runs the model at Q3_K_M using about 17.3 GB, and produces roughly 21.5 tokens per second. 132 cards in total can run it.

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