Kosmos-1
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
- 1 March 2023
- Authors
- Shaohan Huang, Li Dong, Wenhui Wang, Yaru Hao, Saksham Singhal, Shuming Ma, Tengchao Lv, Lei Cui, Owais Khan Mohammed, Barun Patra, Qiang Liu, Kriti Aggarwal, Zewen Chi, Johan Bjorck, Vishrav Chaudhary, Subhojit Som, Xia Song, Furu Wei
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Multimodal, Language, Vision
- Task
- Visual question answering, Image captioning, Language modeling/generation, Chat, Question answering, Document classification, Image classification, Visual puzzles
- Base model
- ViT-G/14 (LiT)
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
- 1.6B
- Training data
- 360,000,000,000 tokens
- Batch size
- 1,200,000
"The total number of parameters of KOSMOS-1 is about 1.6B" "The MLLM component has 24 layers with 2,048 hidden dimensions, 8,192 FFN intermediate size, and 32 attention heads, resulting in about 1.3B parameters."
"We use a batch size of 1.2 million tokens (0.5 million tokens from text corpora, 0.5 million tokens from image-caption pairs, and 0.2 million tokens from interleaved data) and train KOSMOS-1 for 300k steps, corresponding to about 360 billion 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
- 3.5 × 10²¹ FLOP
- How it was established
- Operation counting
6 FLOP / parameter / token * 1.6 * 10^9 parameters [assuming all parameters were updated each step including vision encoder] * 360 * 10^9 tokens = 3.456e+21 FLOP "Likely" confidence because the number of trained parameters could be slightly less
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
MIT license https://github.com/microsoft/unilm/tree/master/kosmos-1 (repo refers to code and weights of Kosmos-2, not Kosmos-1)
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Likely
Sources
Where this record came from and when it was last checked.
- Reference
- Language Is Not All You Need: Aligning Perception with Language Models
- Last updated
- 28 November 2025
What the numbers mean
About this model
Kosmos-1 was published by Microsoft, in United States of America, in March 2023. industry is the category the publisher falls under.
It works in Multimodal, Language, Vision, and is recorded as doing visual question answering, Image captioning, Language modeling/generation, Chat, Question answering, Document classification, Image classification, Visual puzzles.
It builds on ViT-G/14 (LiT), which is why it shares that model's general shape and size.
Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.
Training and provenance
The training run consumed about 3.5 × 10²¹ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
The training set ran to roughly 360,000,000,000 tokens.
Answers
Kosmos-1 — common questions
How much compute was used to train Kosmos-1?
Around 3.5 × 10²¹ FLOP. 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.
What GPU do I need to run Kosmos-1?
None. Kosmos-1 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.
Is Kosmos-1 open source?
No. Kosmos-1 has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does Kosmos-1 have?
Kosmos-1 has 1.6B parameters. "The total number of parameters of KOSMOS-1 is about 1.6B" "The MLLM component has 24 layers with 2,048 hidden dimensions, 8,192 FFN intermediate size, and 32 attention heads, resulting in about 1.3B 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.
Who created Kosmos-1?
Kosmos-1 was published by Microsoft, based in United States of America, categorised as industry.
When was Kosmos-1 released?
Kosmos-1 was published in March 2023. 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 Kosmos-1 used for?
Kosmos-1 works in Multimodal, Language, Vision, and is recorded as handling visual question answering, Image captioning, Language modeling/generation, Chat, Question answering, Document classification, Image classification, Visual puzzles. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
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.