PaLM (540B)
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
- 4 April 2022
- Authors
- Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, Parker Schuh, Kensen Shi, Sasha Tsvyashchenko, Joshua Maynez, Abhishek Rao, Parker Barnes, Yi Tay, Noam Shazeer, Vinodkumar Prabhakaran, Emily Reif, Nan Du, Ben Hutchinson, Reiner Pope, James Bradbury, Jacob Austin, Michael Isard, Guy Gur-A…
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling, Code generation, Translation
- Approach
- Self-supervised learning
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
- 540.4B
- Training data
- 780,000,000,000 tokens
- Epochs
- 1
- Batch size
- 4,000,000
"To further our understanding of the impact of scale on few-shot learning, we trained a 540-billion parameter, densely activated, Transformer language model, which we call Pathways Language Model (PaLM)."
"The PaLM pretraining dataset consists of a high-quality corpus of 780 billion tokens that represent a wide range of natural language use cases." 1 token ~ 0.75 words
"For the largest model, we use batch size 512 (1M tokens) until step 50k, then double it to 1024 (2M tokens) until step 115k, and finally double again it to 2048 (4M tokens) until training is complete at step 255k"
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
- 2.5 × 10²⁴ FLOP
- How it was established
- Hardware
See Table 20. 6144 TPUv4 for 1200 hours + 3072 TPUv4 for 336 hours. Equivalent to 6144 TPUv4 for 1368 hours. 46.2% model FLOPs utilization "The 540B-parameter PaLM model sustained a remarkable 57.8% of the peak hardware floating point performance over 50 days while training on TPU v4 supercomputers." https://cloud.google.com/blog/topics/systems/tpu-v4-enables-performance-energy-and-co2e-efficiency-gains
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
- Google TPU v4
- Chips used
- 6,144
- Chip-hours
- 8,404,992
- Wall-clock time
- 1,536 hours (64 days)
- Hardware utilisation
- MFU 46.2% · HFU 57.8%
- Power draw
- 4.2 MW
- Compute cost
- $3,060,365
6144 TPUv4 for 1200 hours + 3072 TPUv4 for 336 hours. Equivalent to 6144 TPUv4 for 1368 hours.
"In Section 4, we describe how we were able to scale pipeline-free training of PaLM 540B to 6144 chips across two TPU v4 Pods while achieving very high efficiency of 46.2% in model FLOPs utilization (observed throughput relative to theoretical max throughput) and 57.8% in hardware FLOPs utilization." MFU = 0.4620 HFU = 0.5780
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.
- Frontier model
- Yes
- Foundation model
- Yes
- Likely above 10²³ FLOP
- Yes
- Why it is tracked
- Highly cited,SOTA improvement,Training cost
- Record confidence
- Confident
- Citations
- 7,999
Table 4 Demonstrates continued benefits of scaling, as well as discontinuous improvements in performance
Sources
Where this record came from and when it was last checked.
- Reference
- PaLM: Scaling Language Modeling with Pathways
- Last updated
- 25 May 2026
What the numbers mean
Where it came from
PaLM (540B) was published by Google Research, in the country recorded as United States of America, during April 2022. It comes out of an organisation categorised as industry.
It works in the domain of Language, and is recorded as performing the task of language modeling, Code generation, Translation.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
What went into building it
Producing it required arithmetic totalling around 2.5 × 10²⁴ FLOP, on hardware recorded as Google TPU v4. 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 780,000,000,000 tokens of text.
The reason it appears in this catalogue at all: highly cited,SOTA improvement,Training cost.
Answers
PaLM (540B) — common questions
PaLM (540B)— is it open source?
No. Its weights have not been published, so it exists only as a service controlled by its owner.
PaLM (540B)— how many parameters does it have?
It has a parameter count of 540.4B. "To further our understanding of the impact of scale on few-shot learning, we trained a 540-billion parameter, densely activated, Transformer language model, which we call Pathways Language Model (PaLM).". 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.
PaLM (540B)— who created it?
It was published by Google Research, based in United States of America, an organisation categorised as industry.
PaLM (540B)— when was it released?
It was published in April 2022. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.
PaLM (540B)— what is it used for?
It works in the domain of Language, and is recorded as handling the task of language modeling, Code generation, Translation. These are the areas it was designed around; they describe intent rather than a hard boundary.
PaLM (540B)— how much compute was used to train it?
Training consumed around 2.5 × 10²⁴ FLOP, on hardware recorded as Google TPU v4. 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.
PaLM (540B)— 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.
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.