ERNIE-4.5-VL-424B-A47B (文心大模型4.5) TPS calculator

Open weights Baidu 424B parameters March 2025

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

4 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Radeon Instinct MI325X

256 GB · IQ4_XS · 63.8 tok/s

Fastest card

B300

102 tok/s · 288 GB

Which GPUs can run ERNIE-4.5-VL-424B-A47B (文心大模型4.5)?

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.

4 cards match

Calculating
Needs Quantisation Fit
102 tok/s

62–164 · low confidence

B300 NVIDIA 288 GB 8,000 GB/s Sep 2025 235.4 GB Q4_K_M Tight
81.8 tok/s

49–131 · low confidence

Radeon Instinct MI350X AMD 288 GB 8,190 GB/s Jan 2025 235.4 GB Q4_K_M Tight
81.8 tok/s

49–131 · low confidence

Radeon Instinct MI355X AMD 288 GB 8,190 GB/s Jan 2025 235.4 GB Q4_K_M Tight
63.8 tok/s

38–102 · low confidence

Radeon Instinct MI325X AMD 256 GB 6,000 GB/s Oct 2024 210.8 GB IQ4_XS 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
Baidu
Organisation type
Industry
Country
China
Published
16 March 2025

What it does

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

Domain
Multimodal, Language, Vision, Video
Task
Language modeling/generation, Visual question answering, Video description, Speech recognition (ASR), Quantitative reasoning, Code generation, Translation, Question answering, Character recognition (OCR)
Base model
ERNIE-4.5-300B-A47B
Numerical format
FP8

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
424B

MoE: total parameters - 424B active parameters - 47B "ViT encoder comprising 630 million parameters"

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 (unrestricted)
Training code
Unreleased

Apache 2.0: https://huggingface.co/baidu/ERNIE-4.5-VL-424B-A47B-Base-PT Apache 2.0 for inference code https://github.com/PaddlePaddle/ERNIE

Hugging Face
baidu

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
Why it is tracked
SOTA improvement

ERNIE 4.5 surpassed GPT-4o in six out of seven evaluated benchmarks: [Table 7, Table 8] CCBench: Evaluates common-sense reasoning across text and images. ERNIE 4.5 scored approximately 81, slightly outperforming GPT-4o’s ~79. OCRBench: Assesses optical character recognition capabilities, focusing on text extraction from images. ERNIE 4.5 achieved around 88, surpassing GPT-4o’s ~81. ChartQA: Tests understanding of data presented in charts. ERNIE 4.5 scored ~82, marginally ahead of GPT-4o’s ~81. M…

Record confidence
Confident

Sources

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

Reference
Baidu's ERNIE 4.5 & X1: Features, Access, DeepSeek Comparison
Last updated
11 February 2026

The extremes

The ten fastest GPUs that run ERNIE-4.5-VL-424B-A47B (文心大模型4.5)

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 · Q4_K_M 102 tok/s
  2. 02 Radeon Instinct MI350X 288 GB · 8,190 GB/s · Q4_K_M 81.8 tok/s
  3. 03 Radeon Instinct MI355X 288 GB · 8,190 GB/s · Q4_K_M 81.8 tok/s
  4. 04 Radeon Instinct MI325X 256 GB · 6,000 GB/s · IQ4_XS 63.8 tok/s

What the numbers mean

The hardware side

Minimum card

Radeon Instinct MI325X

Memory needed

210.8 GB

Fastest

102 tok/s

ERNIE-4.5-VL-424B-A47B (文心大模型4.5) reaches a parameter count of 424B. That is beyond what any single graphics card holds. Running it means either splitting it across several cards or renting hardware built for the job, and every card able to hold it alone is a datacentre part. The number that can: 4.

At the low end it is handled by Radeon Instinct MI325X, with a memory capacity of 256 GB, running it at a compression of IQ4_XS and producing around 63.8 tokens per second.

The fastest we calculate for it is B300, generating roughly 102 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

What this model is

ERNIE-4.5-VL-424B-A47B (文心大模型4.5) was published by Baidu, in the country recorded as China, during March 2025. The publishing organisation is categorised as industry.

It works in the domain of Multimodal, Language, Vision, Video, and is recorded as performing the task of language modeling/generation, Visual question answering, Video description, Speech recognition (ASR), Quantitative reasoning, Code generation, Translation, Question answering, Character recognition (OCR).

It builds on ERNIE-4.5-300B-A47B. That is why it shares the base model's general shape and size.

Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded. On Hugging Face it is published under the organisation baidu.

What decides the speed

Across every card that can run it, the middle of the range sits at 81.8 tokens per second. Exceeding reading speed outright: 4 of them.

Mixture-of-experts routing is why the speeds here look high for the parameter count. Only a fraction is read per token, but the whole thing has to be loaded.

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.

How it was trained

The reason it appears in this catalogue at all: sOTA improvement.

Step by step

How to choose a GPU for ERNIE-4.5-VL-424B-A47B (文心大模型4.5)

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

  1. 01

    Read the memory figure first

    Start from what it actually needs, which is the requirement of ERNIE-4.5-VL-424B-A47B (文心大模型4.5), needing around 210.8 GB at a compression of IQ4_XS. No amount of processing power compensates for a card that cannot hold it.

  2. 02

    Set the context length you will work at

    Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for ERNIE-4.5-VL-424B-A47B (文心大模型4.5).

  3. 03

    Choose how far you will compress it

    Each card runs the least-compressed copy it can hold, reaching a compression of IQ4_XS 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.

  4. 04

    Compare tokens per second, not specifications

    Ranking by tokens per second follows memory bandwidth rather than core counts, for ERNIE-4.5-VL-424B-A47B (文心大模型4.5). It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B300, at 102 tok/s.

  5. 05

    Look at the headroom, not just the fit

    The fit column separates cards that just manage it from those with room to spare, in the case of ERNIE-4.5-VL-424B-A47B (文心大模型4.5). 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.

  6. 06

    See what else that card runs

    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 ERNIE-4.5-VL-424B-A47B (文心大模型4.5).

Answers

ERNIE-4.5-VL-424B-A47B (文心大模型4.5) — common questions

01

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— how much VRAM does it need?

It needs about 210.8 GB at a compression of IQ4_XS, 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.

02

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— 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.

03

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— how many parameters does it have?

It has a parameter count of 424B. MoE: total parameters - 424B active parameters - 47B "ViT encoder comprising 630 million 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.

04

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— who created it?

It was published by Baidu, based in China, an organisation categorised as industry.

05

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— when was it released?

It was published in March 2025.

06

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— what is it used for?

It works in the domain of Multimodal, Language, Vision, Video, and is recorded as handling the task of language modeling/generation, Visual question answering, Video description, Speech recognition (ASR), Quantitative reasoning, Code generation, Translation, Question answering, Character recognition (OCR). Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

07

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— where can I download it?

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

08

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— can I run it if it does not fit in my GPU?

It can be split between the card and system memory, but it generates painfully slowly that way. The nearest miss we calculate falls short by 62.6 GB. Every figure here assumes the whole model is resident on the card.

09

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— 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: 4. So a second card is rarely the answer here.

10

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— why does the quantisation differ between cards?

Each card is shown running the least-compressed copy it can hold. The number of distinct compression levels across the cards that fit it: 2. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

11

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— how accurate are these speed estimates?

They are calculated from specifications rather than measured, and each carries a range. One example: 62–164 tok/s on B300. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

12

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— what GPU do I need to run it?

The smallest card in our catalogue that holds it is Radeon Instinct MI325X, with a memory capacity of 256 GB. It runs the model at a compression of IQ4_XS using about 210.8 GB, and produces roughly 63.8 tokens per second. The number of cards able to run it in total: 4.

13

ERNIE-4.5-VL-424B-A47B (文心大模型4.5)— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B300, at about 102 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: 4.

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.