PaLI-X
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
- Google Research
- Organisation type
- Industry
- Country
- United States of America
- Published
- 29 May 2023
- Authors
- Xi Chen, Josip Djolonga, Piotr Padlewski, Basil Mustafa, Soravit Changpinyo, Jialin Wu, Carlos Riquelme Ruiz, Sebastian Goodman, Xiao Wang, Yi Tay, Siamak Shakeri, Mostafa Dehghani, Daniel Salz, Mario Lucic, Michael Tschannen, Arsha Nagrani, Hexiang Hu, Mandar Joshi, Bo Pang, Ceslee Montgomery, Paulina Pietrzyk, Marvin Ritter, AJ Piergiovanni, Matthias Minderer, Filip Pavetic, Austin Waters, Gang …
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Multimodal, Language, Vision, Video
- Task
- Image captioning, Video description, Character recognition (OCR), Visual question answering
- Base model
- UL2
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
- 55B
- Training data
- tokens
55B (table 1)
1 billion images with alt texts in WebLI, 400m images in Episodic WebLI data
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
- 5.6 × 10²³ FLOP
"we present PaLI-X [...] consisting of a pretrained large-capacity visual encoder (using [6] as the starting point) and a pretrained language-only encoder-decoder (using [7] as the starting point), further trained at-scale on a vision-and-language data mixture using a combination of self-supervision and full-supervision signals." "Visual component Our visual backbone is scaled to 22B parameters, as introduced by [6], the largest dense ViT model to date." (elsewhere they specify this is ViT-22B)…
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.
- Why it is tracked
- SOTA improvement
- Record confidence
- Likely
- Citations
- 282
"PaLI-X advances the state-of-the-art on most vision-and-language benchmarks considered (25+ of them)." Table 1, Table 2
Sources
Where this record came from and when it was last checked.
- Reference
- PaLI-X: On Scaling up a Multilingual Vision and Language Model
- Last updated
- 25 May 2026
What the numbers mean
Where it came from
PaLI-X was published by Google Research, in United States of America, in May 2023. industry is the category the publisher falls under.
It works in Multimodal, Language, Vision, Video, and is recorded as doing image captioning, Video description, Character recognition (OCR), Visual question answering.
Its starting point was UL2 — most models at this scale are adapted from an existing base rather than built from nothing.
Its weights were never published, so it can only be reached through its provider. No graphics card changes that.
Training and provenance
The training run consumed about 5.6 × 10²³ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
It is tracked in the underlying dataset for one reason in particular: sOTA improvement.
Answers
PaLI-X — common questions
Is PaLI-X open source?
No. PaLI-X has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does PaLI-X have?
PaLI-X has 55B parameters. 55B (table 1). 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 PaLI-X?
PaLI-X was published by Google Research, based in United States of America, categorised as industry.
When was PaLI-X released?
PaLI-X was published in May 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 PaLI-X used for?
PaLI-X works in Multimodal, Language, Vision, Video, and is recorded as handling image captioning, Video description, Character recognition (OCR), Visual question answering. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.
How much compute was used to train PaLI-X?
Around 5.6 × 10²³ 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 PaLI-X?
None. PaLI-X 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.
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.