Our World
in Data
Oxford Martin School logoUniversity of Oxford logoGlobal Change Data Lab logo
Data

Number of large-scale AI systems released per year

See all data and research on:
Artificial Intelligence

What you should know about this indicator

  • Game systems are specifically designed for games and excel in understanding and strategizing gameplay. For instance, AlphaGo, developed by DeepMind, defeated the world champion in the game of Go. Such systems use complex algorithms to compete effectively, even against skilled human players.
  • Language systems are tailored to process language, focusing on understanding, translating, and interacting with human languages. Examples include chatbots, machine translation tools like Google Translate, and sentiment analysis algorithms that can detect emotions in text.
  • Multimodal systems are artificial intelligence frameworks that integrate and interpret more than one type of data input, such as text, images, and audio. ChatGPT-4 is an example of a multimodal model, as it has the capability to process and generate responses based on both textual and visual inputs.
  • Vision systems focus on processing visual information, playing a pivotal role in image recognition and related areas. For example, Facebook's photo tagging model uses vision AI to identify faces.
  • Speech systems are dedicated to handling spoken language, serving as the backbone of voice assistants and similar applications. They recognize, interpret, and generate spoken language to interact with users.
  • Biology systems analyze biological data and simulate biological processes, aiding in drug discovery and genetic research.
  • Image generation systems create visual content from text descriptions or other inputs, used in graphic design and content creation.

How is this data described by its producer?

A foreign key field categorizing the system’s domain of machine learning. This field links to the ML Domains table, and domains are selected from the options in that table.

Number of large-scale AI systems released per year
Describes the specific area, application, or field in which a large-scale AI model is designed to operate. The 2026 data is incomplete and was last updated 07 March 2026.
Source
Epoch AI (2025) – with major processing by Our World in Data
Last updated
March 12, 2025
Next expected update
May 2026
Date range
2019–2026
Unit
AI systems

Sources and processing

This data is based on the following sources

Epoch AI – Tracking Compute-Intensive AI Models

A dataset that tracks compute-intensive AI models, with training compute over 10²³ floating point operations (FLOP). This corresponds to training costs of hundreds of thousands of dollars or more.

To identify compute-intensive AI models, the team at Epoch AI used various resources, estimating compute when not directly reported. They included benchmarks and repositories, such as Papers With Code and Hugging Face, to find models exceeding 10²³ FLOP. They also explored non-English media and specific leaderboards, particularly focusing on Chinese sources.

Additionally, they examined blog posts, press releases from major labs, and scholarly literature to track new models. A separate table was created for models with unconfirmed but plausible compute levels. Despite thorough methods, proprietary and secretive models may have been missed.

Retrieved on
March 7, 2026
Retrieved from
https://epoch.ai/blog/tracking-compute-intensive-ai-models
Citation
This is the citation of the original data obtained from the source, prior to any processing or adaptation by Our World in Data. To cite data downloaded from this page, please use the suggested citation given in Reuse This Work below.
Robi Rahman, David Owen and Josh You (2024), "Tracking Compute-Intensive AI Models". Published online at epochai.org. Retrieved from: 'https://epoch.ai/blog/tracking-compute-intensive-ai-models' [online resource]

A dataset that tracks compute-intensive AI models, with training compute over 10²³ floating point operations (FLOP). This corresponds to training costs of hundreds of thousands of dollars or more.

To identify compute-intensive AI models, the team at Epoch AI used various resources, estimating compute when not directly reported. They included benchmarks and repositories, such as Papers With Code and Hugging Face, to find models exceeding 10²³ FLOP. They also explored non-English media and specific leaderboards, particularly focusing on Chinese sources.

Additionally, they examined blog posts, press releases from major labs, and scholarly literature to track new models. A separate table was created for models with unconfirmed but plausible compute levels. Despite thorough methods, proprietary and secretive models may have been missed.

Retrieved on
March 7, 2026
Retrieved from
https://epoch.ai/blog/tracking-compute-intensive-ai-models
Citation
This is the citation of the original data obtained from the source, prior to any processing or adaptation by Our World in Data. To cite data downloaded from this page, please use the suggested citation given in Reuse This Work below.
Robi Rahman, David Owen and Josh You (2024), "Tracking Compute-Intensive AI Models". Published online at epochai.org. Retrieved from: 'https://epoch.ai/blog/tracking-compute-intensive-ai-models' [online resource]

How we process data at Our World in Data

All data and visualizations on Our World in Data rely on data sourced from one or several original data providers. Preparing this original data involves several processing steps. Depending on the data, this can include standardizing country names and world region definitions, converting units, calculating derived indicators such as per capita measures, as well as adding or adapting metadata such as the name or the description given to an indicator.

At the link below you can find a detailed description of the structure of our data pipeline, including links to all the code used to prepare data across Our World in Data.

Read about our data pipeline
Notes on our processing step for this indicator

The count of large-scale AI models AI systems per domain is derived by tallying the instances of machine learning models classified under each domain category. It's important to note that a single machine learning model can fall under multiple domains. The classification into domains is determined by the specific area, application, or field that the AI model is primarily designed to operate within.

Reuse this work

  • All data produced by third-party providers and made available by Our World in Data are subject to the license terms from the original providers. Our work would not be possible without the data providers we rely on, so we ask you to always cite them appropriately (see below). This is crucial to allow data providers to continue doing their work, enhancing, maintaining and updating valuable data.
  • All data, visualizations, and code produced by Our World in Data are completely open access under the Creative Commons BY license. You have the permission to use, distribute, and reproduce these in any medium, provided the source and authors are credited.

Citations

How to cite this page

To cite this page overall, including any descriptions, FAQs or explanations of the data authored by Our World in Data, please use the following citation:

“Data Page: Number of large-scale AI systems released per year”, part of the following publication: Charlie Giattino, Edouard Mathieu, Veronika Samborska, and Max Roser (2023) - “Artificial Intelligence”. Data adapted from Epoch AI. Retrieved from https://archive.ourworldindata.org/20260308-063423/grapher/number-of-large-scale-ai-systems-released-per-year.html [online resource] (archived on March 8, 2026).

How to cite this data

In-line citationIf you have limited space (e.g. in data visualizations), you can use this abbreviated in-line citation:

Epoch AI (2025) – with major processing by Our World in Data

Full citation

Epoch AI (2025) – with major processing by Our World in Data. “Number of large-scale AI systems released per year” [dataset]. Epoch AI, “Tracking Compute-Intensive AI Models” [original data]. Retrieved April 1, 2026 from https://archive.ourworldindata.org/20260308-063423/grapher/number-of-large-scale-ai-systems-released-per-year.html (archived on March 8, 2026).

Download

Quick download

Download the data shown in this chart as a ZIP file containing a CSV file, metadata in JSON format, and a README. The CSV file can be opened in Excel, Google Sheets, and other data analysis tools.

Download full data

Includes all entities and time points

Download displayed data

Includes only the entities and time points currently visible in the chart

Data API

Use these URLs to programmatically access this chart's data and configure your requests with the options below. Our documentation provides more information on how to use the API, and you can find a few code examples below.

Data URL (CSV format)
https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.csv?v=1&csvType=full&useColumnShortNames=false
Metadata URL (JSON format)
https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.metadata.json?v=1&csvType=full&useColumnShortNames=false

Code examples

Examples of how to load this data into different data analysis tools.

Excel / Google Sheets
=IMPORTDATA("https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.csv?v=1&csvType=full&useColumnShortNames=false")
Python with Pandas
import pandas as pd
import requests

# Fetch the data.
df = pd.read_csv("https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.csv?v=1&csvType=full&useColumnShortNames=false", storage_options = {'User-Agent': 'Our World In Data data fetch/1.0'})

# Fetch the metadata
metadata = requests.get("https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.metadata.json?v=1&csvType=full&useColumnShortNames=false").json()
R
library(jsonlite)

# Fetch the data
df <- read.csv("https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.csv?v=1&csvType=full&useColumnShortNames=false")

# Fetch the metadata
metadata <- fromJSON("https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.metadata.json?v=1&csvType=full&useColumnShortNames=false")
Stata
import delimited "https://ourworldindata.org/grapher/number-of-large-scale-ai-systems-released-per-year.csv?v=1&csvType=full&useColumnShortNames=false", encoding("utf-8") clear