Florence

Closed weights Microsoft 893M parameters November 2021

No estimate

No hardware requirements for this model

The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.

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
Microsoft
Organisation type
Industry
Country
United States of America
Published
22 November 2021
Authors
Lu Yuan, Dongdong Chen, Yi-Ling Chen, Noel Codella, Xiyang Dai, Jianfeng Gao, Houdong Hu, Xuedong Huang, Boxin Li, Chunyuan Li, Ce Liu, Mengchen Liu, Zicheng Liu, Yumao Lu, Yu Shi, Lijuan Wang, JianFeng Wang, Bin Xiao, Zhen Xiao, Jianwei Yang, Michael Zeng, Luowei Zhou, Pengchuan Zhang

What it does

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

Domain
Vision
Task
Image captioning, Visual question answering, Image classification, Object detection
Approach
Supervised
Numerical format
FP16

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
893M

"Our Florence pretrained model has in total 893M parameters, including the language transformer with 256M parameters and the CoSwin-H transformer with 637M parameters."

Training data
7,500,000,000 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
4.8 × 10²² FLOP

"The model takes 10 days to train on 512 NVIDIA A100 GPUs with 40GB memory per GPU." 512 * 312 teraFLOPS * 10 days * 35% utilization = 4.831e22 FLOP

How it was established
Hardware

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 40 GB
Chips used
512
Chip-hours
122,880
Wall-clock time
240 hours (10 days)

10 days on 512 A100 40GB

Power draw
412.8 kW
Compute cost
$106,951

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
Closed — provider access only
Model access
Unreleased
Training code
Unreleased

How it is classified

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

Foundation model
Yes
Why it is tracked
Historical significance,SOTA improvement

"Florence achieves new state-of-the-art results in majority of 44 representative benchmarks, e.g., ImageNet-1K zero-shot classification with top-1 accuracy of 83.74 and the top-5 accuracy of 97.18, 62.4 mAP on COCO fine tuning, 80.36 on VQA, and 87.8 on Kinetics-600." "image retrieval zero-shot (90.9/76.7 R@1 on Flickr30K image-to-text / text-to-image, 64.7/47.2 R@1 on MSCOCO image-to-text / text-to-image) and fine-tuning (97.2/87.9 R@1 on Flickr30K image-to-text / text-to-image, 81.8/63.2 R@1 …

Record confidence
Confident
Citations
1,122

Sources

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

Reference
Florence: A New Foundation Model for Computer Vision
Last updated
25 May 2026

What the numbers mean

Background

Florence was published by Microsoft, in the country recorded as United States of America, during November 2021. It comes out of an organisation categorised as industry.

It works in the domain of Vision, and is recorded as performing the task of image captioning, Visual question answering, Image classification, Object detection.

Its weights were never published, so it can only be reached through its provider. No graphics card changes that.

Training and provenance

Producing it required arithmetic totalling around 4.8 × 10²² FLOP, on hardware recorded as NVIDIA A100 SXM4 40 GB. That figure measures what producing the model cost, and has no bearing on how fast it answers.

It was trained on a corpus of about 7,500,000,000 tokens of text.

The reason it appears in this catalogue at all: historical significance,SOTA improvement.

Answers

Florence — common questions

01

Florence— who created it?

It was published by Microsoft, based in United States of America, an organisation categorised as industry.

02

Florence— when was it released?

It was published in November 2021. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

03

Florence— what is it used for?

It works in the domain of Vision, and is recorded as handling the task of image captioning, Visual question answering, Image classification, Object detection. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

04

Florence— how much compute was used to train it?

Training consumed around 4.8 × 10²² FLOP, on hardware recorded as NVIDIA A100 SXM4 40 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.

05

Florence— what GPU do I need to run it?

None. This is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.

06

Florence— is it open source?

No. Its weights have not been published, so it exists only as a service controlled by its owner.

07

Florence— how many parameters does it have?

It has a parameter count of 893M. "Our Florence pretrained model has in total 893M parameters, including the language transformer with 256M parameters and the CoSwin-H transformer with 637M 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.

Source

Original publication

Record last updated 25 May 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.