Med-PaLM
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,National Library of Medicine,DeepMind
- Organisation type
- Industry,Government,Industry
- Country
- United States of America, United Kingdom of Great Britain and Northern Ireland
- Published
- 12 July 2023
- Authors
- Karan Singhal, Shekoofeh Azizi, Tao Tu, S. Sara Mahdavi, Jason Wei, Hyung Won Chung, Nathan Scales, Ajay Tanwani, Heather Cole-Lewis, Stephen Pfohl, Perry Payne, Martin Seneviratne, Paul Gamble, Chris Kelly, Abubakr Babiker, Nathanael Schärli, Aakanksha Chowdhery, Philip Mansfield, Dina Demner-Fushman, Blaise Agüera y Arcas, Dale Webster, Greg S. Corrado, Yossi Matias, Katherine Chou, Juraj Gottwe…
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Medicine, Language
- Task
- Question answering
- Base model
- Flan-PaLM 540B
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
- 540B
- Training data
- tokens
"We performed instruction prompt tuning on Flan-PaLM 540B with a soft prompt length of 100 to produce Med-PaLM. We froze the rest of the model, and used an embedding dimension of 18432 as in PaLM [1], which resulted in 1.84M trainable parameters." (from supplementary materials)
from supplementary materials "We used a batch size of 32 across all runs and ran training for 200 steps." they also mention average question length of 25 words, but I am not sure if that applies to all datapoints MedMCQA (https://proceedings.mlr.press/v174/pal22a/pal22a.pdf, Table 2) has on average 12.77+ 2.69+67.52 = 82.98 tokens per datapoint
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.
- Record confidence
- Confident
Sources
Where this record came from and when it was last checked.
- Reference
- Large language models encode clinical knowledge
- Last updated
- 28 November 2025
What the numbers mean
What this model is
Med-PaLM was published by Google Research,National Library of Medicine,DeepMind, in United States of America, in July 2023. The organisation is categorised as industry,Government,Industry.
It works in Medicine, Language, and is recorded as doing question answering.
Its starting point was Flan-PaLM 540B — most models at this scale are adapted from an existing base rather than built from nothing.
Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.
Answers
Med-PaLM — common questions
Is Med-PaLM open source?
No. Med-PaLM has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does Med-PaLM have?
Med-PaLM has 540B parameters. "We performed instruction prompt tuning on Flan-PaLM 540B with a soft prompt length of 100 to produce Med-PaLM. We froze the rest of the model, and used an embedding dimension of 18432 as in PaLM [1], which resulted in 1.84M trainable parameters." (from supplementary materials). 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 Med-PaLM?
Med-PaLM was published by Google Research,National Library of Medicine,DeepMind, based in United States of America, categorised as industry,Government,Industry.
When was Med-PaLM released?
Med-PaLM was published in July 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 Med-PaLM used for?
Med-PaLM works in Medicine, Language, and is recorded as handling question answering. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
What GPU do I need to run Med-PaLM?
None. Med-PaLM 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.