---
title: "AI Timeline"
type: timeline
id: "timeline"
description: "Chronological history of artificial intelligence from 1950 to 2026 — key research breakthroughs, product launches, milestones, and policy events."
last_updated: "2026-04-10"
---

# AI Timeline

A chronological history of artificial intelligence from 1950 to 2026. Each event is typed by category (Research, Product, Policy) and scope (Major, Minor) — structured temporal data that agents can filter and query via the [JSON API](/api/v1/timeline.json).

A chronological history of artificial intelligence from 1950 to 2026.

## The Birth of AI (1950–1969)

*Visionaries laid the theoretical groundwork and built the first AI programs, fueled by optimism about machine intelligence.*

### 1950 — Turing Test Proposed

*Research (Major)*

Alan Turing publishes "Computing Machinery and Intelligence," proposing the imitation game as a test for machine intelligence.

### 1956-08 — Dartmouth Conference

*Milestone (Major)*

John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon organize the Dartmouth workshop, coining the term "artificial intelligence."

### 1958 — Perceptron Invented

*Research*

Frank Rosenblatt builds the Mark I Perceptron at Cornell, the first hardware implementation of an artificial neural network.

### 1966 — ELIZA Chatbot

*Product (Major)*

Joseph Weizenbaum at MIT creates ELIZA, an early natural language processing program that simulates a Rogerian psychotherapist.

### 1969 — Shakey the Robot

*Research*

SRI International develops Shakey, the first general-purpose mobile robot able to reason about its own actions.

## AI Winters & Expert Systems (1970–1993)

*Funding cuts and unmet expectations led to "AI winters," but expert systems and neural network research kept the field alive.*

### 1970 — First AI Winter Begins

*Milestone*

The Lighthill Report and DARPA funding cuts trigger the first AI winter, as early systems fail to meet lofty expectations.

### 1980 — Expert Systems Boom

*Product*

Rule-based expert systems like XCON (R1) at DEC demonstrate commercial viability, sparking a new wave of AI investment.

### 1986 — Backpropagation Popularized

*Research (Major)*

Rumelhart, Hinton, and Williams publish their influential paper on backpropagation, enabling practical training of multi-layer neural networks.

### 1988 — Second AI Winter

*Milestone*

The collapse of the LISP machine market and limitations of expert systems trigger a second wave of reduced AI funding.

## The AI Renaissance (1997–2011)

*Increased computing power and data availability produced landmark achievements, from chess to Jeopardy!.*

### 1997-05 — Deep Blue Defeats Kasparov

*Milestone (Major)*

IBM's Deep Blue defeats world chess champion Garry Kasparov in a six-game match, marking a milestone for AI in strategic games.

### 2002 — Roomba Released

*Product*

iRobot releases the Roomba, one of the first commercially successful autonomous robots, bringing AI into millions of homes.

### 2006 — Geoffrey Hinton Coins "Deep Learning"

*Research (Major)*

Hinton and collaborators demonstrate that deep belief networks can be efficiently trained layer by layer, reigniting interest in neural networks.

### 2011-02 — IBM Watson Wins Jeopardy!

*Milestone (Major)*

IBM Watson defeats Jeopardy! champions Ken Jennings and Brad Rutter, showcasing advances in natural language understanding.

### 2011-10 — Siri Launches

*Product*

Apple introduces Siri with the iPhone 4S, bringing AI-powered voice assistants to mainstream consumer devices.

## The Deep Learning Revolution (2012–2019)

*Deep neural networks shattered benchmarks in vision, language, and games, ushering in the modern AI era.*

### 2012-09 — AlexNet Wins ImageNet

*Research (Major)*

Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton's deep CNN wins the ImageNet competition by a huge margin, sparking the deep learning revolution.

### 2014-06 — GANs Introduced

*Research (Major)*

Ian Goodfellow and colleagues introduce Generative Adversarial Networks (GANs), enabling AI to generate realistic images and data.

### 2014-09 — Amazon Alexa Launches

*Product*

Amazon launches Alexa and the Echo speaker, accelerating the AI voice assistant market and smart home adoption.

### 2016-03 — AlphaGo Defeats Lee Sedol

*Milestone (Major)*

DeepMind's AlphaGo defeats Go world champion Lee Sedol 4-1, a landmark for AI in games once thought beyond machine capability.

### 2017-06 — Transformer Architecture Published

*Research (Major)*

Google researchers publish "Attention Is All You Need," introducing the Transformer architecture that becomes the foundation of modern AI.

### 2018-06 — GPT-1 Released

*Research*

OpenAI releases GPT-1, demonstrating that generative pre-training on large text corpora followed by fine-tuning achieves strong NLP results.

### 2018-10 — BERT Released

*Research (Major)*

Google releases BERT (Bidirectional Encoder Representations from Transformers), significantly advancing language understanding benchmarks.

### 2019-02 — GPT-2 Released

*Product*

OpenAI releases GPT-2, initially withholding the full model due to concerns about misuse, sparking debate about responsible AI release.

## The Generative AI Era (2020–2024)

*Large language models, image generators, and multimodal AI went mainstream, transforming how billions of people work and create.*

### 2020-06 — GPT-3 Released

*Product (Major)*

OpenAI releases GPT-3 with 175 billion parameters, demonstrating remarkable few-shot learning and text generation capabilities.

### 2020-11 — AlphaFold 2 Solves Protein Folding

*Milestone (Major)*

DeepMind's AlphaFold 2 achieves near-experimental accuracy in protein structure prediction at CASP14, a breakthrough for biology.

### 2021-01 — DALL-E Announced

*Product (Major)*

OpenAI reveals DALL-E, a model that generates images from text descriptions, demonstrating creative multimodal AI capabilities.

### 2021-06 — GitHub Copilot Launched

*Product (Major)*

GitHub launches Copilot, an AI pair programmer powered by OpenAI Codex, fundamentally changing how developers write code.

### 2022-04 — DALL-E 2 Released

*Product*

OpenAI releases DALL-E 2 with dramatically improved image quality, making AI art generation accessible to a wider audience.

### 2022-08 — Stable Diffusion Released

*Product (Major)*

Stability AI releases Stable Diffusion as open source, democratizing AI image generation and sparking a creative revolution.

### 2022-11 — ChatGPT Launches

*Product (Major)*

OpenAI releases ChatGPT, reaching 100 million users in two months and bringing conversational AI into the global mainstream.

### 2023-02 — Bing Chat & Google Bard

*Product*

Microsoft integrates ChatGPT into Bing, and Google announces Bard, triggering an AI arms race among tech giants.

### 2023-03 — GPT-4 Released

*Product (Major)*

OpenAI releases GPT-4, a multimodal large language model showing significant improvements in reasoning, creativity, and accuracy.

### 2023-07 — Claude 2 & Llama 2 Released

*Product (Major)*

Anthropic releases Claude 2 with a 100K context window, and Meta open-sources Llama 2, expanding the AI ecosystem.

### 2023-10 — Biden Executive Order on AI

*Policy (Major)*

President Biden signs a sweeping executive order establishing safety standards, privacy protections, and equity guidelines for AI.

### 2023-12 — Gemini Launched

*Product (Major)*

Google DeepMind launches Gemini, a natively multimodal model designed to compete at the frontier of AI capabilities.

### 2024-02 — Gemini 1.5 Pro

*Product*

Google releases Gemini 1.5 Pro with a 1-million-token context window, pushing the boundaries of long-context understanding.

### 2024-03 — Claude 3 Family Released

*Product (Major)*

Anthropic releases the Claude 3 family (Haiku, Sonnet, Opus), with Opus setting new benchmarks across reasoning and analysis tasks.

### 2024-04 — Meta Llama 3 Released

*Product*

Meta releases Llama 3 in 8B and 70B sizes, raising the bar for open-source language models.

### 2024-05 — GPT-4o Released

*Product (Major)*

OpenAI releases GPT-4o ("omni"), a natively multimodal model processing text, audio, and vision with faster response times.

### 2024-07 — EU AI Act Enters into Force

*Policy (Major)*

The European Union AI Act, the world's first comprehensive AI regulation framework, officially enters into force.

### 2024-09 — OpenAI o1 Released

*Product (Major)*

OpenAI introduces o1, a model trained with reinforcement learning to perform complex reasoning using chain-of-thought before responding.

### 2024-12 — Sora Video Generation Preview

*Product*

OpenAI opens access to Sora, an AI model that generates realistic videos from text prompts, advancing generative media.

## The Agentic AI Era (2025–2026)

*AI models gained autonomous agency, using tools, writing and reviewing code, and completing multi-step workflows. Rapid model releases, consolidation in the AI IDE market, and sweeping regulation defined a new phase of the AI revolution.*

### 2025-01 — DeepSeek R1 Released

*Product (Major)*

Chinese AI lab DeepSeek releases R1, an open-source reasoning model rivaling frontier models at a fraction of the training cost.

### 2025-02 — Claude 3.7 Sonnet Released

*Product*

Anthropic releases Claude 3.7 Sonnet, introducing a hybrid reasoning mode that combines standard and extended thinking.

### 2025-03 — GPT-4.5 Released

*Product*

OpenAI releases GPT-4.5, a large model focused on improved creativity, reduced hallucinations, and broader world knowledge.

### 2025-03 — Gemini 2.5 Pro Released

*Product*

Google DeepMind releases Gemini 2.5 Pro, a thinking model with strong reasoning and a 1-million-token context window.

### 2025-04 — Llama 4 Released

*Product (Major)*

Meta releases Llama 4 Scout and Maverick, mixture-of-experts models with a 10-million-token context window.

### 2025-05 — Claude Opus 4 & Sonnet 4 Released

*Product (Major)*

Anthropic releases Claude Opus 4 (classified Level 3 on their safety scale) and Claude Sonnet 4. Opus 4 becomes the most capable Claude model yet.

### 2025-05 — DeepSeek R1-0528 Update

*Product*

DeepSeek updates R1 to version R1-0528 with improved reasoning capabilities, while R2 remains delayed.

### 2025-07 — Grok 4 and Grok 4 Heavy Launch

*Product (Major)*

xAI launches Grok 4 and Grok 4 Heavy, claiming Grok 4 is 'the most intelligent model in the world.' Includes native tool use and real-time search.

### 2025-07 — Amazon Kiro IDE Preview

*Product*

Amazon launches Kiro, a spec-driven AI coding IDE that generates specifications before writing code. Runs on Amazon Bedrock.

### 2025-07 — Cognition AI Acquires Windsurf

*Milestone*

Cognition AI (makers of Devin) acquires Windsurf AI IDE, consolidating the AI coding tools market.

### 2025-08 — GPT-5 Launches

*Product (Major)*

OpenAI releases GPT-5, a unified multimodal system with smart routing between standard and thinking modes. Sets new benchmarks: 94.6% on AIME, 74.9% on SWE-bench. 256K token context window.

### 2025-08 — Claude Opus 4.1 Released

*Product*

Anthropic releases Opus 4.1 with improved code generation, search reasoning, and instruction adherence.

### 2025-08 — EU AI Act GPAI Rules Take Effect

*Policy (Major)*

Governance rules and obligations for General-Purpose AI models become applicable under the EU AI Act.

### 2025-08 — DeepSeek V3.1 Released

*Product*

DeepSeek releases V3.1 under MIT License with hybrid thinking/non-thinking modes, surpassing prior models by 40% on SWE-bench.

### 2025-09 — Claude Sonnet 4.5 Released

*Product*

Anthropic releases Claude Sonnet 4.5 achieving 77.2% on SWE-bench Verified.

### 2025-09 — DeepSeek V3.2 Released

*Product (Major)*

DeepSeek releases V3.2 with sparse attention mechanism, performing on par with GPT-5.1 and Gemini 3 Pro on benchmarks.

### 2025-11 — Google Gemini 3 Pro & Deep Think Launch

*Product (Major)*

Google releases Gemini 3 Pro and 3 Deep Think, replacing Gemini 2.5 series. State-of-the-art reasoning with multimodal capabilities.

### 2025-11 — Google Antigravity AI IDE Launches Free

*Product*

Google launches Antigravity, a free agent-first AI IDE with built-in Claude Opus 4.6 and Gemini 3.1 Pro.

### 2025-11 — xAI Grok 4.1 Released

*Product*

xAI releases Grok 4.1 with 65% fewer hallucinations (down to 4.22%), available in Thinking and Non-thinking configurations.

### 2025-11 — Claude Opus 4.5 Released

*Product*

Anthropic releases Opus 4.5 with a 67% price cut, making premium intelligence more affordable.

### 2025-12 — Mistral 3 Model Suite Released

*Product*

Mistral releases their new flagship model suite with Apache 2.0 license, adopting MoE architecture. European alternative for enterprise AI.

### 2025-12 — Gemini 3 Flash Released

*Product*

Google releases Gemini 3 Flash, replacing Gemini 2.5 Flash as default model in consumer app and Google Search AI responses.

### 2025-12 — Trump AI Executive Order

*Policy (Major)*

White House issues executive order 'Ensuring a National Policy Framework for AI', creating an AI Litigation Task Force to challenge state AI laws. Aims to preempt state-by-state regulation.

### 2025-12 — GPT-5.2 Released

*Product*

OpenAI releases GPT-5.2, reportedly hastened by Google's Gemini 3 launch.

### 2026-01 — Apple Partners with Google Gemini for Siri

*Milestone (Major)*

Apple announces plans to use Gemini AI in the next version of Siri, marking a major partnership shift.

### 2026-01 — ChatGPT Go Plan Launches Globally

*Product*

OpenAI launches the $8/month Go plan worldwide, sitting between Free and Plus tiers.

### 2026-01 — Multiple State AI Laws Take Effect

*Policy*

California AI Safety Act, California Transparency in Frontier AI Act, and Texas Responsible AI Governance Act take effect January 1.

### 2026-02 — Claude Opus 4.6 Released

*Product (Major)*

Anthropic releases Claude Opus 4.6 with 1M token context (no surcharge), agent teams, and PowerPoint integration. Scores 80.8% on SWE-bench, leading all models.

### 2026-02 — Claude Sonnet 4.6 Released

*Product*

First Sonnet model preferred over previous Opus generation in coding evaluations. 1M token context at $3/$15 per million tokens.

### 2026-02 — Gemini 3.1 Pro Released

*Product*

Google releases Gemini 3.1 Pro with improved reasoning and agentic capabilities.

### 2026-02 — EU AI Act Prohibited Practices Enforced

*Policy (Major)*

EU begins enforcing prohibited AI practices including social scoring and real-time biometric surveillance. AI literacy obligations also take effect.

### 2026-02 — Ads Appear in Free ChatGPT

*Milestone*

OpenAI begins showing advertisements to Free and Go tier ChatGPT users in the US.

### 2026-03 — GPT-5.4 Released

*Product (Major)*

OpenAI releases GPT-5.4, combining frontier reasoning, coding (GPT-5.3-Codex capabilities), and agentic workflows into a single model.

### 2026-03 — OpenAI Shuts Down Sora

*Milestone (Major)*

OpenAI discontinues Sora video generation tool on March 25, pivoting resources toward robotics and world simulation. Underwhelming commercial returns and fierce competition cited.

### 2026-03 — Claude Code Source Code Leak

*Milestone (Major)*

Anthropic accidentally ships 512,000 lines of Claude Code source code via npm packaging error. Second security incident in a week after CMS misconfiguration exposed 'Claude Mythos' model details.

### 2026-03 — GitHub Copilot Agentic Code Review Ships

*Product*

GitHub launches Agentic Code Review on March 5, expanding Copilot from code completion to autonomous code review.

AI Timeline

A chronological history of artificial intelligence from 1950 to 2026. Each event is typed by category (Research, Product, Policy) and scope (Major, Minor) — structured temporal data that agents can filter and query via the JSON API.

A chronological history of artificial intelligence from 1950 to 2026.

The Birth of AI (1950–1969)

Visionaries laid the theoretical groundwork and built the first AI programs, fueled by optimism about machine intelligence.

1950 — Turing Test Proposed

Research (Major)

Alan Turing publishes "Computing Machinery and Intelligence," proposing the imitation game as a test for machine intelligence.

1956-08 — Dartmouth Conference

Milestone (Major)

John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon organize the Dartmouth workshop, coining the term "artificial intelligence."

1958 — Perceptron Invented

Research

Frank Rosenblatt builds the Mark I Perceptron at Cornell, the first hardware implementation of an artificial neural network.

1966 — ELIZA Chatbot

Product (Major)

Joseph Weizenbaum at MIT creates ELIZA, an early natural language processing program that simulates a Rogerian psychotherapist.

1969 — Shakey the Robot

Research

SRI International develops Shakey, the first general-purpose mobile robot able to reason about its own actions.

AI Winters & Expert Systems (1970–1993)

Funding cuts and unmet expectations led to "AI winters," but expert systems and neural network research kept the field alive.

1970 — First AI Winter Begins

Milestone

The Lighthill Report and DARPA funding cuts trigger the first AI winter, as early systems fail to meet lofty expectations.

1980 — Expert Systems Boom

Product

Rule-based expert systems like XCON (R1) at DEC demonstrate commercial viability, sparking a new wave of AI investment.

1986 — Backpropagation Popularized

Research (Major)

Rumelhart, Hinton, and Williams publish their influential paper on backpropagation, enabling practical training of multi-layer neural networks.

1988 — Second AI Winter

Milestone

The collapse of the LISP machine market and limitations of expert systems trigger a second wave of reduced AI funding.

The AI Renaissance (1997–2011)

Increased computing power and data availability produced landmark achievements, from chess to Jeopardy!.

1997-05 — Deep Blue Defeats Kasparov

Milestone (Major)

IBM's Deep Blue defeats world chess champion Garry Kasparov in a six-game match, marking a milestone for AI in strategic games.

2002 — Roomba Released

Product

iRobot releases the Roomba, one of the first commercially successful autonomous robots, bringing AI into millions of homes.

2006 — Geoffrey Hinton Coins "Deep Learning"

Research (Major)

Hinton and collaborators demonstrate that deep belief networks can be efficiently trained layer by layer, reigniting interest in neural networks.

2011-02 — IBM Watson Wins Jeopardy!

Milestone (Major)

IBM Watson defeats Jeopardy! champions Ken Jennings and Brad Rutter, showcasing advances in natural language understanding.

2011-10 — Siri Launches

Product

Apple introduces Siri with the iPhone 4S, bringing AI-powered voice assistants to mainstream consumer devices.

The Deep Learning Revolution (2012–2019)

Deep neural networks shattered benchmarks in vision, language, and games, ushering in the modern AI era.

2012-09 — AlexNet Wins ImageNet

Research (Major)

Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton's deep CNN wins the ImageNet competition by a huge margin, sparking the deep learning revolution.

2014-06 — GANs Introduced

Research (Major)

Ian Goodfellow and colleagues introduce Generative Adversarial Networks (GANs), enabling AI to generate realistic images and data.

2014-09 — Amazon Alexa Launches

Product

Amazon launches Alexa and the Echo speaker, accelerating the AI voice assistant market and smart home adoption.

2016-03 — AlphaGo Defeats Lee Sedol

Milestone (Major)

DeepMind's AlphaGo defeats Go world champion Lee Sedol 4-1, a landmark for AI in games once thought beyond machine capability.

2017-06 — Transformer Architecture Published

Research (Major)

Google researchers publish "Attention Is All You Need," introducing the Transformer architecture that becomes the foundation of modern AI.

2018-06 — GPT-1 Released

Research

OpenAI releases GPT-1, demonstrating that generative pre-training on large text corpora followed by fine-tuning achieves strong NLP results.

2018-10 — BERT Released

Research (Major)

Google releases BERT (Bidirectional Encoder Representations from Transformers), significantly advancing language understanding benchmarks.

2019-02 — GPT-2 Released

Product

OpenAI releases GPT-2, initially withholding the full model due to concerns about misuse, sparking debate about responsible AI release.

The Generative AI Era (2020–2024)

Large language models, image generators, and multimodal AI went mainstream, transforming how billions of people work and create.

2020-06 — GPT-3 Released

Product (Major)

OpenAI releases GPT-3 with 175 billion parameters, demonstrating remarkable few-shot learning and text generation capabilities.

2020-11 — AlphaFold 2 Solves Protein Folding

Milestone (Major)

DeepMind's AlphaFold 2 achieves near-experimental accuracy in protein structure prediction at CASP14, a breakthrough for biology.

2021-01 — DALL-E Announced

Product (Major)

OpenAI reveals DALL-E, a model that generates images from text descriptions, demonstrating creative multimodal AI capabilities.

2021-06 — GitHub Copilot Launched

Product (Major)

GitHub launches Copilot, an AI pair programmer powered by OpenAI Codex, fundamentally changing how developers write code.

2022-04 — DALL-E 2 Released

Product

OpenAI releases DALL-E 2 with dramatically improved image quality, making AI art generation accessible to a wider audience.

2022-08 — Stable Diffusion Released

Product (Major)

Stability AI releases Stable Diffusion as open source, democratizing AI image generation and sparking a creative revolution.

2022-11 — ChatGPT Launches

Product (Major)

OpenAI releases ChatGPT, reaching 100 million users in two months and bringing conversational AI into the global mainstream.

2023-02 — Bing Chat & Google Bard

Product

Microsoft integrates ChatGPT into Bing, and Google announces Bard, triggering an AI arms race among tech giants.

2023-03 — GPT-4 Released

Product (Major)

OpenAI releases GPT-4, a multimodal large language model showing significant improvements in reasoning, creativity, and accuracy.

2023-07 — Claude 2 & Llama 2 Released

Product (Major)

Anthropic releases Claude 2 with a 100K context window, and Meta open-sources Llama 2, expanding the AI ecosystem.

2023-10 — Biden Executive Order on AI

Policy (Major)

President Biden signs a sweeping executive order establishing safety standards, privacy protections, and equity guidelines for AI.

2023-12 — Gemini Launched

Product (Major)

Google DeepMind launches Gemini, a natively multimodal model designed to compete at the frontier of AI capabilities.

2024-02 — Gemini 1.5 Pro

Product

Google releases Gemini 1.5 Pro with a 1-million-token context window, pushing the boundaries of long-context understanding.

2024-03 — Claude 3 Family Released

Product (Major)

Anthropic releases the Claude 3 family (Haiku, Sonnet, Opus), with Opus setting new benchmarks across reasoning and analysis tasks.

2024-04 — Meta Llama 3 Released

Product

Meta releases Llama 3 in 8B and 70B sizes, raising the bar for open-source language models.

2024-05 — GPT-4o Released

Product (Major)

OpenAI releases GPT-4o ("omni"), a natively multimodal model processing text, audio, and vision with faster response times.

2024-07 — EU AI Act Enters into Force

Policy (Major)

The European Union AI Act, the world's first comprehensive AI regulation framework, officially enters into force.

2024-09 — OpenAI o1 Released

Product (Major)

OpenAI introduces o1, a model trained with reinforcement learning to perform complex reasoning using chain-of-thought before responding.

2024-12 — Sora Video Generation Preview

Product

OpenAI opens access to Sora, an AI model that generates realistic videos from text prompts, advancing generative media.

The Agentic AI Era (2025–2026)

AI models gained autonomous agency, using tools, writing and reviewing code, and completing multi-step workflows. Rapid model releases, consolidation in the AI IDE market, and sweeping regulation defined a new phase of the AI revolution.

2025-01 — DeepSeek R1 Released

Product (Major)

Chinese AI lab DeepSeek releases R1, an open-source reasoning model rivaling frontier models at a fraction of the training cost.

2025-02 — Claude 3.7 Sonnet Released

Product

Anthropic releases Claude 3.7 Sonnet, introducing a hybrid reasoning mode that combines standard and extended thinking.

2025-03 — GPT-4.5 Released

Product

OpenAI releases GPT-4.5, a large model focused on improved creativity, reduced hallucinations, and broader world knowledge.

2025-03 — Gemini 2.5 Pro Released

Product

Google DeepMind releases Gemini 2.5 Pro, a thinking model with strong reasoning and a 1-million-token context window.

2025-04 — Llama 4 Released

Product (Major)

Meta releases Llama 4 Scout and Maverick, mixture-of-experts models with a 10-million-token context window.

2025-05 — Claude Opus 4 & Sonnet 4 Released

Product (Major)

Anthropic releases Claude Opus 4 (classified Level 3 on their safety scale) and Claude Sonnet 4. Opus 4 becomes the most capable Claude model yet.

2025-05 — DeepSeek R1-0528 Update

Product

DeepSeek updates R1 to version R1-0528 with improved reasoning capabilities, while R2 remains delayed.

2025-07 — Grok 4 and Grok 4 Heavy Launch

Product (Major)

xAI launches Grok 4 and Grok 4 Heavy, claiming Grok 4 is 'the most intelligent model in the world.' Includes native tool use and real-time search.

2025-07 — Amazon Kiro IDE Preview

Product

Amazon launches Kiro, a spec-driven AI coding IDE that generates specifications before writing code. Runs on Amazon Bedrock.

2025-07 — Cognition AI Acquires Windsurf

Milestone

Cognition AI (makers of Devin) acquires Windsurf AI IDE, consolidating the AI coding tools market.

2025-08 — GPT-5 Launches

Product (Major)

OpenAI releases GPT-5, a unified multimodal system with smart routing between standard and thinking modes. Sets new benchmarks: 94.6% on AIME, 74.9% on SWE-bench. 256K token context window.

2025-08 — Claude Opus 4.1 Released

Product

Anthropic releases Opus 4.1 with improved code generation, search reasoning, and instruction adherence.

2025-08 — EU AI Act GPAI Rules Take Effect

Policy (Major)

Governance rules and obligations for General-Purpose AI models become applicable under the EU AI Act.

2025-08 — DeepSeek V3.1 Released

Product

DeepSeek releases V3.1 under MIT License with hybrid thinking/non-thinking modes, surpassing prior models by 40% on SWE-bench.

2025-09 — Claude Sonnet 4.5 Released

Product

Anthropic releases Claude Sonnet 4.5 achieving 77.2% on SWE-bench Verified.

2025-09 — DeepSeek V3.2 Released

Product (Major)

DeepSeek releases V3.2 with sparse attention mechanism, performing on par with GPT-5.1 and Gemini 3 Pro on benchmarks.

2025-11 — Google Gemini 3 Pro & Deep Think Launch

Product (Major)

Google releases Gemini 3 Pro and 3 Deep Think, replacing Gemini 2.5 series. State-of-the-art reasoning with multimodal capabilities.

2025-11 — Google Antigravity AI IDE Launches Free

Product

Google launches Antigravity, a free agent-first AI IDE with built-in Claude Opus 4.6 and Gemini 3.1 Pro.

2025-11 — xAI Grok 4.1 Released

Product

xAI releases Grok 4.1 with 65% fewer hallucinations (down to 4.22%), available in Thinking and Non-thinking configurations.

2025-11 — Claude Opus 4.5 Released

Product

Anthropic releases Opus 4.5 with a 67% price cut, making premium intelligence more affordable.

2025-12 — Mistral 3 Model Suite Released

Product

Mistral releases their new flagship model suite with Apache 2.0 license, adopting MoE architecture. European alternative for enterprise AI.

2025-12 — Gemini 3 Flash Released

Product

Google releases Gemini 3 Flash, replacing Gemini 2.5 Flash as default model in consumer app and Google Search AI responses.

2025-12 — Trump AI Executive Order

Policy (Major)

White House issues executive order 'Ensuring a National Policy Framework for AI', creating an AI Litigation Task Force to challenge state AI laws. Aims to preempt state-by-state regulation.

2025-12 — GPT-5.2 Released

Product

OpenAI releases GPT-5.2, reportedly hastened by Google's Gemini 3 launch.

2026-01 — Apple Partners with Google Gemini for Siri

Milestone (Major)

Apple announces plans to use Gemini AI in the next version of Siri, marking a major partnership shift.

2026-01 — ChatGPT Go Plan Launches Globally

Product

OpenAI launches the $8/month Go plan worldwide, sitting between Free and Plus tiers.

2026-01 — Multiple State AI Laws Take Effect

Policy

California AI Safety Act, California Transparency in Frontier AI Act, and Texas Responsible AI Governance Act take effect January 1.

2026-02 — Claude Opus 4.6 Released

Product (Major)

Anthropic releases Claude Opus 4.6 with 1M token context (no surcharge), agent teams, and PowerPoint integration. Scores 80.8% on SWE-bench, leading all models.

2026-02 — Claude Sonnet 4.6 Released

Product

First Sonnet model preferred over previous Opus generation in coding evaluations. 1M token context at $3/$15 per million tokens.

2026-02 — Gemini 3.1 Pro Released

Product

Google releases Gemini 3.1 Pro with improved reasoning and agentic capabilities.

2026-02 — EU AI Act Prohibited Practices Enforced

Policy (Major)

EU begins enforcing prohibited AI practices including social scoring and real-time biometric surveillance. AI literacy obligations also take effect.

2026-02 — Ads Appear in Free ChatGPT

Milestone

OpenAI begins showing advertisements to Free and Go tier ChatGPT users in the US.

2026-03 — GPT-5.4 Released

Product (Major)

OpenAI releases GPT-5.4, combining frontier reasoning, coding (GPT-5.3-Codex capabilities), and agentic workflows into a single model.

2026-03 — OpenAI Shuts Down Sora

Milestone (Major)

OpenAI discontinues Sora video generation tool on March 25, pivoting resources toward robotics and world simulation. Underwhelming commercial returns and fierce competition cited.

2026-03 — Claude Code Source Code Leak

Milestone (Major)

Anthropic accidentally ships 512,000 lines of Claude Code source code via npm packaging error. Second security incident in a week after CMS misconfiguration exposed 'Claude Mythos' model details.

2026-03 — GitHub Copilot Agentic Code Review Ships

Product

GitHub launches Agentic Code Review on March 5, expanding Copilot from code completion to autonomous code review.