AI Timeline

ChatGPT launches as a public research preview

OpenAI opens ChatGPT to the public and makes conversational AI a mainstream product category overnight.

Alphabet loses about $100 billion after Bard demo error

Google's rushed Bard rollout backfires when a demo answer proves inaccurate, sharpening investor anxiety around the AI race ChatGPT ignited.

Caused by OpenAI: ChatGPT launches as a public research preview

GPT-4 becomes OpenAI's new flagship model

GPT-4 raises the bar on reasoning, reliability, and multimodal potential for commercial AI systems.

IBM pauses hiring for roles it expects AI to automate

IBM says it will slow or suspend hiring in back-office functions where generative AI is expected to absorb routine work over time.

Caused by OpenAI: GPT-4 becomes OpenAI's new flagship model

Chegg stock collapses after warning ChatGPT is hurting growth

Chegg becomes one of the clearest early public-market casualties of generative AI after telling investors ChatGPT is weakening new customer growth.

Caused by OpenAI: ChatGPT launches as a public research preview

Nvidia erupts after AI demand rewrites its revenue outlook

Nvidia's blowout forecast confirms that the generative AI boom is translating directly into infrastructure spending and market value.

Caused by OpenAI: GPT-4 becomes OpenAI's new flagship model

Claude 2 arrives with longer context and public access

Anthropic brings Claude 2 to both API users and the public-facing claude.ai experience.

ChatGPT adds voice and image understanding

OpenAI turns ChatGPT into a more multimodal assistant with voice conversations and image input.

OpenAI launches custom GPTs

Users and teams can package instructions, tools, and knowledge into reusable GPT-based products.

Google introduces Gemini 1.0

Gemini launches as Google's first natively multimodal model family across Ultra, Pro, and Nano.

Duolingo cuts contractors as AI takes on more content work

Duolingo confirms it reduced its contractor workforce as GPT-4-era tooling accelerated translation and content production.

Caused by OpenAI: GPT-4 becomes OpenAI's new flagship model

Gemini 1.5 Pro pushes long-context frontier

Google DeepMind unveils Gemini 1.5 with a major leap in multimodal performance and context length.

Sora previews OpenAI's text-to-video model

OpenAI shows a major jump in generative video quality and simulation-like scene consistency.

Klarna says its AI assistant is doing the work of 700 agents

Klarna reports that its OpenAI-powered assistant is now handling most customer service chats and matching the output of a large human support team.

Caused by OpenAI: ChatGPT launches as a public research preview

Anthropic launches the Claude 3 model family

Claude 3 Haiku, Sonnet, and Opus position Anthropic as a top-tier frontier model provider.

Meta AI expands globally on top of Llama 3

Meta turns Llama 3 into a consumer assistant layer across Facebook, Instagram, WhatsApp, Messenger, and the web.

Google shows Project Astra and launches Gemini 1.5 Flash

Google pairs a faster Gemini model with a clearer vision for always-on, agent-like assistants.

Mistral releases Codestral for software work

Mistral expands into code generation with a specialized model aimed at developer workflows.

Qwen2 broadens Alibaba's open model push

Qwen2 expands the family with stronger multilingual, coding, and long-context performance across multiple sizes.

Claude 3.5 Sonnet sets a new baseline for coding assistants

Anthropic ships its strongest everyday model and widens Claude access across plans and API surfaces.

GPT-4o mini makes low-cost multimodal AI practical

OpenAI introduces a cheaper high-quality model for lightweight assistants, apps, and automation.

Meta ships Llama 3.1 and the 405B open model push

Llama 3.1 strengthens Meta's open-model strategy and makes frontier-scale open weights more credible.

DeepSeek-V2.5 merges chat and coding into one open model

DeepSeek unifies its general and coder lines into a stronger all-in-one release for web and API use.

OpenAI debuts o1-preview reasoning models

OpenAI formally introduces a reasoning-first line designed to spend more time thinking before answering.

Pixtral 12B brings multimodal vision to Mistral's lineup

Mistral adds its first multimodal model with image understanding and document-friendly workflows.

Qwen2.5 becomes one of the biggest open model releases yet

Alibaba expands Qwen with a large family of general, coding, and math models aimed at broad developer adoption.

Llama 3.2 expands Meta's open vision and edge strategy

Meta pushes Llama further onto devices and into vision use cases with a broader open model portfolio.

Anthropic launches computer use in public beta

Claude starts interacting with graphical interfaces, moving frontier models closer to true digital agents.

ChatGPT search arrives with live web answers

OpenAI turns ChatGPT into a real-time answer engine with citations and web-native retrieval.

Palantir hits a record as Wall Street rewards AI monetization

Palantir's raised outlook and stronger AI platform demand push the stock to a fresh record as investors look for enterprise AI winners.

Caused by OpenAI: GPT-4 becomes OpenAI's new flagship model

Gemini 2.0 marks Google's shift into the agentic era

Gemini 2.0 Flash launches alongside deeper agent prototypes, multimodal outputs, and new developer APIs.

DeepSeek-V3 pushes open-weight MoE models into the top tier

DeepSeek releases a faster and more capable MoE flagship, narrowing the gap with leading closed models.

MiniMax-01 debuts with ultra-long context and open weights

MiniMax releases its 01 series with Lightning Attention, a 4 million token context window, and multimodal support.

DeepSeek-R1 makes reasoning-first open models mainstream

DeepSeek launches an open reasoning model positioned against OpenAI o1, with web, API, and MIT-licensed weights.

Operator previews OpenAI's web-using agent

OpenAI productizes browser-based task execution and pushes consumer-facing agent workflows forward.

Qwen2.5-Max scales Alibaba's frontier MoE ambitions

Qwen2.5-Max extends the Qwen line with a much larger MoE system aimed at top-tier chat and coding benchmarks.

Deep research brings analyst-style web workflows to ChatGPT

OpenAI introduces long-running, source-heavy research tasks as a first-class ChatGPT capability.

Claude 3.7 Sonnet and Claude Code land together

Anthropic pairs a hybrid reasoning model with a terminal-native coding workflow for developers.

GPT-4.5 becomes OpenAI's biggest chat model

GPT-4.5 emphasizes broader world knowledge, improved chat quality, and lower hallucination rates.

Gemma 3 strengthens Google's open-model story

Google upgrades its lightweight open model family for single-GPU and edge-oriented deployment.

Gemini 2.5 Pro becomes Google's new reasoning flagship

Google introduces a thinking-first Gemini release aimed at harder coding, science, and planning tasks.

4o image generation turns GPT-4o into a stronger visual tool

OpenAI folds high-fidelity image generation directly into GPT-4o and ChatGPT workflows.

GPT-4.1 targets developers with longer context and better coding

OpenAI launches a new API-focused GPT family optimized for instruction following and code-heavy workloads.

Qwen3 adds hybrid reasoning modes to the Qwen family

Alibaba introduces Qwen3 with thinking and non-thinking modes, stronger agent behavior, and broad open-weight coverage.

Meta launches a standalone Meta AI app built on Llama 4

Meta reframes its assistant as a dedicated product with voice, memory, and a stronger cross-device story.

Chegg lays off 22% of staff as AI shakes up edtech demand

Chegg announces a deeper restructuring after sustained pressure from AI-powered study tools and weaker demand for its legacy homework-help business.

Caused by OpenAI: ChatGPT launches as a public research preview

Claude 4 becomes Anthropic's new workhorse generation

Anthropic advances its production line again, with Claude 4 positioned as the stronger everyday agent model.

Mistral introduces Magistral as its first reasoning model

Magistral expands Mistral beyond general-purpose assistants into transparent, domain-heavy reasoning tasks.

MiniMax-M1 brings large-scale open reasoning to MiniMax

MiniMax follows its long-context work with a hybrid-attention reasoning model tuned for productivity-heavy tasks.

ChatGPT agent merges research and action

OpenAI combines browsing, tools, and execution into a broader agent mode for real-world tasks.

Amazon moves to cut thousands of manager roles as AI reshapes office work

Reports on Amazon's management reduction tie the cuts to AI-enabled efficiency and a broader push to flatten the company as generative tools take on more coordination work.

Trigger: AI efficiency push and lower management ratios

Gemini 3 starts a new Google model generation

Google positions Gemini 3 as its most capable general-purpose model family for reasoning and multimodality.

Palantir jumps as AI demand drives another guidance raise

Palantir's results reinforce that enterprise and government AI demand is still translating into software revenue and investor enthusiasm.

Trigger: Raised AI demand guidance and commercial momentum

GPT-5.3-Codex expands agentic coding into broader computer work

OpenAI pushes Codex beyond code generation toward longer-running, multi-step professional workflows.

AI software sector suffers a scare-trade selloff

Investors dump richly valued AI software names as fears grow around margins, spending discipline, and how durable the current wave of AI monetization will be.

Trigger: AI scare trade around spending and competitive pressure

Claude Sonnet 4.6 sharpens Anthropic's agent and coding lead

Anthropic refreshes Sonnet again with stronger coding, computer use, and large-context performance.

Gemini 3.1 Pro raises the ceiling for complex reasoning work

Google extends the Gemini 3 line with a stronger Pro model aimed at higher-difficulty planning and analysis.

Nano Banana 2 upgrades Google's image generation stack

Google introduces a faster image model that blends stronger world knowledge with production-friendly editing.

Block cuts nearly 1,000 jobs in an AI-first efficiency push

Jack Dorsey's restructuring at Block is framed around faster execution, fewer layers, and a heavier reliance on AI inside the company's operations.

Trigger: AI-first efficiency and flatter org design

Block shares jump as Wall Street backs the AI-led cuts

Investors respond positively to Block's sharper cost discipline, treating the AI-linked restructuring as a margin and execution story rather than a demand warning.

Trigger: AI-first cost cuts and margin expectations

GPT-5.4 becomes OpenAI's new frontier work model

OpenAI combines reasoning, coding, and computer use into a more capable flagship for professional workflows.

Oracle rallies as AI cloud demand rewrites its backlog story

Oracle's latest results convince investors that AI infrastructure demand is flowing through to backlog, capex, and future revenue at scale.

Trigger: AI cloud revenue surge and backlog forecast

Atlassian cuts 1,600 jobs as Jira parent pivots to an AI-first model

Atlassian says it is cutting about 10% of staff as it reshapes its skill mix and operating model around an AI-first future for products like Jira, Confluence, and Trello.

Trigger: AI-first company transition and product automation shift

Reports say Meta is planning sweeping layoffs as AI costs mount

Reuters reported that Meta is considering broad workforce cuts as the company absorbs heavy AI infrastructure spending, making this a planned layoff item until the reductions are confirmed.

Trigger: Massive AI infrastructure spending and cost pressure