IMPALA

Closed weights DeepMind 1.6M parameters February 2018

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
DeepMind
Organisation type
Industry
Country
United Kingdom of Great Britain and Northern Ireland
Published
5 February 2018
Authors
Lasse Espeholt, Hubert Soyer, Remi Munos, Karen Simonyan, Volodymir Mnih, Tom Ward, Yotam Doron, Vlad Firoiu, Tim Harley, Iain Dunning, Shane Legg, Koray Kavukcuoglu

What it does

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

Domain
Games
Task
Atari
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
1.6M

"Figure 3 in the paper states that the large architecture has 1.6 million parameters. I am using the large model because it was the only one trained on all the Atari games at once, which seems like the most impressive task in the suite." Source: https://docs.google.com/spreadsheets/d/1Kj4Q5WADcDXtUJLIOfGTCE3tGvxNczEMwyy8QtgSkHk/edit#gid=54587040&fvid=1361937389

Training data
11,400,000,000 tokens

From fig 6, there were 1e10 environment frames, and 24 agents. Thus we note down 2.4e11 for the "dataset size"

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
1.7 × 10²⁰ FLOP

Source: Ajeya Cotra and Tom Davidson, https://docs.google.com/spreadsheets/d/1Kj4Q5WADcDXtUJLIOfGTCE3tGvxNczEMwyy8QtgSkHk/edit#gid=54587040&fvid=1361937389

How it was established
Third-party estimation

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 P100
Chips used
1
Chip-hours
100
Wall-clock time
100 hours

Maximum training time for IMPALA is 100 hours according to Figure 6. This seems to refer to the 1 GPU model. The 8 GPU model looks to have been trained about 1/8 as long.

Power draw
286 W
Compute cost
$53

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
Open source

training code, Apache license: https://github.com/google-deepmind/scalable_agent

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
Highly cited,SOTA improvement

not an absolute SOTA "IMPALA is able to achieve better performance than previous agents with less data" "We demonstrate the effectiveness of IMPALA for multi-task reinforcement learning on DMLab-30 (a set of 30 tasks from the DeepMind Lab environment (Beattie et al., 2016)) and Atari-57 (all available Atari games in Arcade Learning Environment (Bellemare et al., 2013a)). Our results show that IMPALA is able to achieve better performance than previous agents with less data"

Record confidence
Confident
Citations
1,829

Sources

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

Reference
IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
Last updated
25 May 2026

What the numbers mean

About this model

IMPALA was published by DeepMind, in United Kingdom of Great Britain and Northern Ireland, in February 2018. industry is the category the publisher falls under.

It works in Games, and is recorded as doing atari.

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

What went into building it

The training run consumed about 1.7 × 10²⁰ FLOP, on NVIDIA P100. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

It was trained on about 11,400,000,000 tokens of text.

Its inclusion criterion is highly cited,SOTA improvement.

Answers

IMPALA — common questions

01

Who created IMPALA?

IMPALA was published by DeepMind, based in United Kingdom of Great Britain and Northern Ireland, categorised as industry.

02

When was IMPALA released?

IMPALA was published in February 2018. 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

What is IMPALA used for?

IMPALA works in Games, and is recorded as handling atari. These are the areas it was designed around; they describe intent rather than a hard boundary.

04

How much compute was used to train IMPALA?

Around 1.7 × 10²⁰ FLOP, on NVIDIA P100. 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

What GPU do I need to run IMPALA?

None. IMPALA 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

Is IMPALA open source?

No. IMPALA has not had its weights published, so it exists only as a service controlled by its owner.

07

How many parameters does IMPALA have?

IMPALA has 1.6M parameters. "Figure 3 in the paper states that the large architecture has 1.6 million parameters. I am using the large model because it was the only one trained on all the Atari games at once, which seems like the most impressive task in the suite." Source: https://docs.google.com/spreadsheets/d/1Kj4Q5WADcDXtUJLIOfGTCE3tGvxNczEMwyy8QtgSkHk/edit#gid=54587040&fvid=1361937389. 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.