· Release
Google releases Gemini 4
Google released Gemini 4 Argon, the first model of the Gemini 4 generation. On October 28 it announced Gemini 4 Carbon, a new top model.
Source Google blog — Gemini 4 Argon
AI Encyclopedia › Timelines by field 1/5
AI timeline
Key moments, newest first
The big moments from 1950, when people first asked whether machines can think, to today, when chat AI has become part of daily life.
- Release 13
- Milestone 7
- Research 3
- Policy 2
2026
· Release
OpenAI releases GPT-6 Astra
OpenAI released GPT-6 Astra, the first model of the GPT-6 generation. GPT-6 Sol and Luna followed on September 22.
Source OpenAI — GPT-6 Astra
· Release
Anthropic releases Claude Sonnet 5
Anthropic released Claude Sonnet 5, a new generation of its model. Claude Opus 5 followed on July 24.
Source Anthropic — Claude Sonnet 5
· Policy
South Korea's AI Basic Act takes effect
The AI Basic Act took effect. It covers labeling of AI-made results, management of "high-impact AI" that can greatly affect life and basic rights, and safety duties for businesses.
2025
· Release
Google announces Gemini 3
Google announced Gemini 3, a new generation of its model.
Source Google blog — Gemini 3
· Release
OpenAI launches GPT-5
OpenAI released GPT-5, which chooses between a fast answer and deep reasoning to fit the question.
Source Wikipedia — GPT-5
· Milestone
AI reaches gold-medal level at the International Mathematical Olympiad
Google DeepMind's Gemini Deep Think solved the problems within the contest's time limit and got a score at the gold-medal level in the organizers' grading.
Source Google DeepMind blog
· Release
Claude 4 and Claude Code launch
With Claude 4, Anthropic made Claude Code, a coding agent that reads and edits code directly on a developer's computer, available to everyone.
· Release
DeepSeek releases R1
China's DeepSeek released its reasoning model R1 together with its weights, and it drew great attention around the world.
Source Wikipedia — DeepSeek
2024
· Milestone
Nobel Prize in Chemistry honors AlphaFold
Demis Hassabis and John Jumper, whose AlphaFold predicted protein structures, and David Baker, who designed new proteins, won the Nobel Prize in Chemistry.
Source Nobel Prize announcement
· Milestone
Nobel Prize in Physics honors artificial neural networks
John Hopfield and Geoffrey Hinton won the Nobel Prize in Physics for their work on artificial neural networks, the base of today's machine learning.
Source Nobel Prize announcement
· Release
OpenAI o1 thinks before it answers
OpenAI released o1, a reasoning model that thinks through several steps before answering. The way AI solves math and coding problems changed.
Source Wikipedia — OpenAI o1
· Policy
The EU Artificial Intelligence Act enters into force
The European Union's Artificial Intelligence Act, which regulates AI by level of risk, took effect. Its rules start to apply step by step over several years.
2023
· Release
Google announces Gemini
Google announced Gemini, a model built from the start to handle text, photos, sound and video together.
· Release
Llama 2 is released for business use too
Meta released Llama 2, with model weights that can be downloaded and used commercially, and open models gained momentum.
· Release
GPT-4 arrives, and the first Claude the same day
OpenAI released GPT-4, which takes in and understands text and photos together. On the same day, Anthropic released its first Claude.
Source Wikipedia — GPT-4
2022
· Release
ChatGPT launches
OpenAI released ChatGPT, which anyone can try by chatting, and chat AI became part of daily life.
Source OpenAI — Introducing ChatGPT
· Release
Stable Diffusion puts image AI in everyone's hands
A model that draws pictures from a text description was released with its weights, and image-generating AI spread quickly.
Source Wikipedia — Stable Diffusion
2020
· Milestone
AlphaFold 2 predicts protein structures
In the CASP14 protein structure prediction contest, it made predictions close to the structures found by experiments, opening a way to solve a 50-year-old problem in biology.
Source Google DeepMind blog
2017
· Research
The Transformer paper is published
The paper "Attention Is All You Need" by Google researchers introduced the Transformer, the structure behind most of today's chat AI.
2016
· Milestone
AlphaGo beats Lee Sedol
Google DeepMind's Go AI AlphaGo won a five-game match in Seoul against 9-dan player Lee Sedol, 4 wins to 1.
2012
· Milestone
AlexNet starts the deep learning wave
The deep neural network AlexNet beat other methods by a wide margin in the ImageNet contest, where programs name the objects in photos. Deep learning research grew a lot after this.
Source Wikipedia — AlexNet
1997
· Milestone
Deep Blue beats the world chess champion
IBM's chess computer Deep Blue won a six-game match against world champion Garry Kasparov.
1956
· Research
"Artificial intelligence" becomes the name of a research field
The Dartmouth summer workshop, organized by John McCarthy and others, was the first research gathering to use the name "artificial intelligence". It is often counted as the start of AI research.
1950
· Research
Turing asks whether machines can think
In his paper "Computing Machinery and Intelligence", Alan Turing proposed a test that tells a person from a machine by conversation alone, now called the Turing test.
The events were chosen and described by AI, and each date was checked against the source listed with it (an official announcement, a paper or Wikipedia). Dates follow the source: Korean events in Korean time, others in local time or UTC, so a day's difference is possible. The timeline does not grow by itself; new events are checked and added by hand. Report an error · Last updated