AI Timeline: 2020–2026

OpenAI launched GPT-3 with 175B parameters, showing that scale alone could produce remarkably capable language models. It sparked the modern LLM era.

OpenAI revealed DALL-E, a model that could generate images from text descriptions. The first mainstream text-to-image system.

GitHub and OpenAI released Copilot, the first AI coding assistant to reach mainstream adoption. It changed how developers write code.

DALL-E 2 brought photorealistic image generation to the public, kicking off the generative art movement.

Meta open-sourced OPT-175B, signaling Big Tech's shift toward open AI research.

Stability AI released Stable Diffusion as open source, democratizing image generation and spawning thousands of community projects.

OpenAI released ChatGPT, reaching 100M users in two months. It became the fastest-growing consumer app in history and brought AI into the mainstream.

Microsoft integrated GPT-4 into Bing Search, firing the first shot in the AI search wars.

Meta released LLaMA, proving smaller open-source models could match larger proprietary ones. It ignited the open-source AI movement.

OpenAI launched GPT-4, a multimodal model that set new records across every major benchmark and passed the bar exam.

Anthropic released Claude, its first public model, emphasizing safety and helpfulness. It became the main GPT-4 competitor.

UK government announced the first global AI Safety Summit, putting AI governance on the world stage.

Meta released Llama 2 with a commercial license, making powerful open-source models accessible to businesses.

President Biden signed the most sweeping US executive order on AI, requiring safety testing for powerful models.

OpenAI launched GPTs (custom chatbots) and GPT-4 Turbo with 128K context, plus the Assistants API.

Google released Gemini, its most capable model, claiming it surpassed GPT-4 on multiple benchmarks.

OpenAI previewed Sora, a text-to-video model generating minute-long photorealistic clips. It redefined what was possible in AI video.

Anthropic released the Claude 3 family (Opus, Sonnet, Haiku), with Opus matching GPT-4 on most benchmarks.

Meta released Llama 3 in 8B and 70B sizes, setting new open-source performance records.

OpenAI launched GPT-4o with native multimodal understanding — text, images, audio, and video in a single model at faster speeds.

Anthropic released Claude 3.5 Sonnet, which outperformed GPT-4o on coding and reasoning at a fraction of the cost.

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

OpenAI released o1, the first "reasoning model" that thinks step-by-step before answering. It crushed PhD-level science and math benchmarks.

Google released Gemini 2.0 Flash with native tool use and agentic capabilities built in.

Chinese lab DeepSeek released R1, a reasoning model matching o1 performance at a fraction of the training cost. It triggered a global market selloff in AI chip stocks.

xAI launched Grok 3, trained on a massive Colossus supercluster. It topped multiple benchmarks and brought real competition to the frontier.

Meta released Llama 4 Scout and Maverick with mixture-of-experts architecture and 10M token context windows.

Anthropic launched Claude 4 Opus and Sonnet with extended thinking, improved agentic coding, and memory capabilities.

Google released Gemini 2.5 Pro with a 1M token context window and state-of-the-art coding performance.

OpenAI launched GPT-5 with significantly improved reasoning, reduced hallucinations, and native agent capabilities.

Multiple providers launched production-ready AI agent frameworks. Autonomous AI systems started handling complex multi-step business workflows.

The UN proposed a global AI governance framework, with 40+ nations signing an initial agreement on frontier model safety testing.