Skip to content
Preview · early version, more releases are being added · early version
gotreplaced.si

Every release

The same timeline as a plain list, newest first. For the interactive version, go to the road.

2026

  1. Anthropic launches Claude Sheets and WhiteboardsClaude gets shared spreadsheets and whiteboards for people on paid plans, in beta.Anthropic
  2. Gemini 4 CarbonGoogle's strongest Gemini, Carbon, reaches paying developers and Ultra subscribers after a security trial.Google
  3. Claude Fable 5.2 and Mythos 5.2Anthropic releases Fable 5.2 and Mythos 5.2, its latest top models, faster and cheaper than 5.1.Anthropic
  4. GPT-6.1 TerraOpenAI releases GPT-6.1 Terra, a fast, low-cost model for computer tasks and writing.OpenAI
  5. Gemini 4 ArgonGoogle opens Gemini 4 Argon, its frontier model for real-world coding, to paying developers.Google
  6. Claude Haiku 5.5Anthropic releases Haiku 5.5, its cheapest and fastest small model.Anthropic
  7. GPT-6.1 SolOpenAI upgrades Sol to GPT-6.1, getting close to Astra on coding and work tasks at a fifth of the price.OpenAI
  8. Claude Sonnet 5.5Anthropic releases Sonnet 5.5, a faster and cheaper Sonnet for everyday tasks.Anthropic
  9. Claude Opus 5.5Anthropic releases Opus 5.5, a Claude model it says matches Fable 5.1 on most work at lower cost.Anthropic
  10. GPT-6 Sol and LunaOpenAI releases two cheaper GPT-6 models, Sol and Luna, for everyday work at lower prices.OpenAI
  11. Grok 4.7xAI releases Grok 4.7, a stronger model for coding and office work at the same price as the last one.xAI
  12. GPT-6 AstraOpenAI introduces GPT-6 Astra, its most advanced model, for computer use, coding and professional work.OpenAI
  13. Claude Fable 5.1 and Mythos 5.1Anthropic updates Fable to 5.1, a stronger and cheaper top model, with Mythos 5.1 for vetted users.Anthropic
  14. Grok 4.6xAI releases Grok 4.6, built for AI agents that work on long multi-step tasks.xAI
  15. Claude Opus 5Anthropic releases Opus 5, a Claude model that comes close to Fable 5 at half the price.Anthropic
  16. Grok 4.5xAI releases Grok 4.5, its strongest model so far for coding and handling longer jobs.xAI
  17. GPT-5.6 family (Sol, Terra, Luna)OpenAI releases three GPT-5.6 models: a flagship, an everyday model and a cheaper, lighter one.OpenAI
  18. Muse Spark 1.1Meta upgrades Muse Spark and opens a public preview of its API so developers can use it.Meta
  19. Claude Sonnet 5Anthropic releases Sonnet 5, its most hands-off Sonnet yet, at a lower price than bigger models.Anthropic
  20. Claude Fable 5 and Claude Mythos 5Anthropic releases Fable 5, its strongest Claude for general use, plus a restricted twin for security work.Anthropic
  21. Claude Opus 4.8Anthropic releases Opus 4.8, with more control over effort and big-task tools in Claude Code.Anthropic
  22. Grok Build early betaxAI launches Grok Build, a coding assistant that runs in the terminal, in early beta.xAI
  23. OpenAI discontinues the Sora app and websiteOpenAI switches off the Sora video app and website; the Sora API followed in September 2026.OpenAI
  24. DeepSeek-V4 PreviewDeepSeek previews V4, a new generation of open models that can read very long documents.DeepSeek
  25. GPT-5.5ChatGPT gets a smarter model that can carry out longer, messier tasks on a computer with less guidance.OpenAI
  26. Claude DesignAnthropic launches Claude Design, which makes designs, slides and prototypes from a written request.Anthropic
  27. Claude Opus 4.7Anthropic releases Opus 4.7, an upgrade to Opus, its Claude model line, better at software work.Anthropic
  28. Muse SparkMeta's new superintelligence lab releases Muse Spark, the first of its Muse AI models, in the Meta AI app.Meta
  29. Gemma 4Google releases Gemma 4, open AI models for phones to workstations, free to use commercially.Google
  30. Gemini 3.1 Pro released in previewGoogle releases Gemini 3.1 Pro in preview, a step up in reasoning from Gemini 3 Pro.Google
  31. Claude Opus 4.6Anthropic releases Opus 4.6, a stronger Claude for coding and long-running tasks.Anthropic
  32. Alexa+ available to everyone in the USAmazon's new AI-powered Alexa that holds conversations and takes actions opens to all US customers.Others
  33. Google Project GenieGoogle lets US Ultra subscribers build and explore short interactive worlds from a prompt or image.Google
  34. Grok Imagine APIxAI opens its Grok Imagine tools for making and editing video and images to developers.xAI
  35. Claude Cowork research previewAnthropic's Claude Desktop gets Cowork, an assistant that works on your files and everyday office tasks.Anthropic

2025

  1. GPT-5.2OpenAI releases GPT-5.2, aimed at professional work like spreadsheets, slides and coding.OpenAI
  2. DeepSeek-V3.2DeepSeek releases V3.2 in its app, website and API, with open weights.DeepSeek
  3. Claude Opus 4.5Anthropic releases Opus 4.5, which it calls its best model for coding, agents and computer use.Anthropic
  4. Gemini 3 ProGoogle launches Gemini 3, its newest top model, in the Gemini app, Search and for developers.Google
  5. Google AntigravityGoogle launches Antigravity, a coding tool built around AI agents, free to try.Google
  6. Grok 4.1xAI rolls out Grok 4.1, an updated Grok available to all users on the web, X and mobile apps.xAI
  7. GPT-5.1ChatGPT gets GPT-5.1, a warmer, more conversational update to GPT-5 with more control over tone.OpenAI
  8. ChatGPT Atlas browserOpenAI releases Atlas, a web browser with ChatGPT built in.OpenAI
  9. Claude Haiku 4.5Anthropic releases Haiku 4.5, its small, low-cost Claude model, to all users.Anthropic
  10. Sora 2 and the Sora appOpenAI launches a new video model with sound, and an app for making and sharing AI videos.OpenAI
  11. DeepSeek-V3.2-ExpDeepSeek releases a test model that is cheaper to run on long inputs, and cuts its prices.DeepSeek
  12. Claude in Chrome pilotClaude can work inside the Chrome browser and take actions for you, starting with a small pilot.Anthropic
  13. DeepSeek-V3.1DeepSeek releases V3.1, one model that can answer fast or think step by step, with open weights.DeepSeek
  14. GPT-5ChatGPT's new model for everyone decides when to answer quickly and when to think longer.OpenAI
  15. Claude Opus 4.1Anthropic upgrades Claude Opus to version 4.1, with better coding and research skills.Anthropic
  16. gpt-oss open-weight modelsOpenAI publishes two language models that anyone can download and run on their own computers.OpenAI
  17. ChatGPT agentChatGPT can now do multi-step tasks for you on its own virtual computer.OpenAI
  18. Grok 4xAI releases Grok 4, which can use tools and search the live web on its own.xAI
  19. DeepSeek-R1-0528DeepSeek updates its R1 reasoning model with better results and fewer made-up answers.DeepSeek
  20. Claude Code generally availableAnthropic coding tool Claude Code leaves its test phase and opens to all developers.Anthropic
  21. Claude Opus 4 and Sonnet 4Anthropic releases two new Claude models, with the Sonnet version free for everyone to use.Anthropic
  22. AI Mode rolls out to all US Search usersGoogle opens AI Mode, a chat-style way to search, to everyone in the US with no sign-up needed.Google
  23. Jules coding agent opens in public betaGoogle opens Jules, an AI helper that works on coding tasks in the background, to everyone.Google
  24. Veo 3 and Imagen 4Google launches Veo 3, which makes video clips with sound, and Imagen 4 for images.Google
  25. Codex cloud coding agentOpenAI launches Codex, an AI coding helper that works on several software tasks at once.OpenAI
  26. Meta AI appMeta launches a standalone Meta AI app, built around voice chat and a feed of what others ask.Meta
  27. Qwen3Alibaba releases Qwen3, free AI models that can answer fast or think longer on hard questions.Others
  28. Gemini 2.5 Flash released in previewGoogle releases Gemini 2.5 Flash, a fast model whose thinking can be switched on or off.Google
  29. OpenAI o3 and o4-miniChatGPT gets two reasoning models that think longer and can use tools like search and code on their own.OpenAI
  30. GPT-4.1 released in the APIOpenAI releases GPT-4.1 models for developers, better at coding and following instructions.OpenAI
  31. Llama 4 Scout and MaverickMeta releases two Llama 4 models anyone can download, and previews a much larger one.Meta
  32. Gemini 2.5 ProGoogle releases a Gemini model that thinks before it answers, for paying app users and developers.Google
  33. GPT-4o image generation in ChatGPTChatGPT's default image maker becomes GPT-4o, which renders text and follows detailed prompts accurately.OpenAI
  34. GPT-4.5 research previewOpenAI releases GPT-4.5, a chat model it says feels more natural and understands intent better.OpenAI
  35. Claude 3.7 Sonnet and Claude CodeClaude gets a model that can answer fast or think step by step, plus a coding helper for the terminal.Anthropic
  36. Grok 3xAI releases Grok 3 in beta, a reasoning-focused chatbot model, to paying X subscribers.xAI
  37. Gemini 2.0 Flash generally availableGoogle makes the updated Gemini 2.0 Flash ready for developers to build real apps on.Google
  38. Deep research launched in ChatGPTChatGPT can now spend many minutes searching the web and write up a detailed research report.OpenAI
  39. DeepSeek-R1DeepSeek releases a chatbot model that thinks step by step, free for anyone to download and run.DeepSeek

2024

  1. DeepSeek-V3DeepSeek releases V3, a large AI model published as open-source, with a technical paper.DeepSeek
  2. Veo 2 and Imagen 3 updateGoogle gives a first group of users a new AI video maker, Veo 2, and an updated image maker.Google
  3. Gemini 2.0 Flash (experimental)Google releases Gemini 2.0, a new generation built for AI that can take actions, starting with Flash.Google
  4. SoraOpenAI opens Sora, its tool for making videos from text descriptions, to paying ChatGPT users.OpenAI
  5. Llama 3.3 70BMeta releases Llama 3.3, a 70B text model with downloadable weights, as an update to its Llama 3 family.Meta
  6. OpenAI o1 and ChatGPT ProChatGPT gets the o1 reasoning model, plus a new $200-a-month Pro plan with extra-strong thinking.OpenAI
  7. Model Context Protocol open-sourcedAnthropic publishes MCP, a shared standard for connecting AI assistants to data and apps.Anthropic
  8. ChatGPT searchChatGPT can now search the web and answer with links to its sources.OpenAI
  9. Apple IntelligenceApple's first AI features arrive on newer iPhones, iPads and Macs: writing help and a more chatty Siri.Others
  10. Computer use introduced in public betaClaude can now look at a screen and click and type like a person, in a first public test.Anthropic
  11. Llama 3.2Meta releases Llama 3.2, models with downloadable weights that understand images, plus small ones for phones.Meta
  12. Qwen2.5Alibaba releases a new family of AI models in many sizes that anyone can download.Others
  13. OpenAI o1-previewOpenAI releases a ChatGPT model that takes time to think through hard problems before answering.OpenAI
  14. Gemini Live starts rolling outGemini gets a voice mode for free-flowing spoken conversations, starting on Android phones.Google
  15. Grok-2 betaxAI releases Grok-2 and Grok-2 mini in beta to paying X subscribers.xAI
  16. Llama 3.1 405BMeta releases Llama 3.1 405B, a very large AI model anyone can download, with upgraded smaller versions.Meta
  17. Runway Gen-3 AlphaRunway's video maker gets a new model that makes more realistic, more controllable clips from text prompts.Others
  18. Claude Projects introducedClaude.ai lets Pro and Team users group chats, files and instructions into Projects.Anthropic
  19. Claude 3.5 SonnetAnthropic's Claude chatbot gets a new model, free to use on the web and the iPhone app.Anthropic
  20. Copilot+ PCs go on saleMicrosoft launches a new kind of Windows laptop built to run AI features on the device itself.Others
  21. AI Overviews begins rolling out in US SearchGoogle Search starts showing AI-written summaries at the top of results for people in the US.Google
  22. GPT-4oOpenAI shows GPT-4o, which can see, hear and talk in real time; ChatGPT gets it now, with voice to follow.OpenAI
  23. AlphaFold 3Google DeepMind unveils AlphaFold 3, an AI that predicts how proteins, DNA and drug molecules fit together.Google
  24. Llama 3Meta releases Llama 3, a new AI model in two sizes that people can download and use.Meta
  25. Gemini 1.5 Pro released in public previewGoogle opens Gemini 1.5 Pro to developers in 180+ countries, after a limited preview in February.Google
  26. Grok-1 open-sourcedxAI publishes the raw Grok-1 model so anyone can download and use it.xAI
  27. Claude 3 familyAnthropic launches Claude 3, a new set of chatbot models, with one free to use on claude.ai.Anthropic
  28. Gemma open modelsGoogle releases Gemma, small AI models that anyone can download and build on.Google
  29. Bard renamed Gemini, Gemini AdvancedGoogle renames its Bard chatbot Gemini and adds a paid plan with its most capable model.Google

2023

  1. Mixtral 8x7BMistral AI releases a free-to-download AI model that anyone can run and build on.Others
  2. NotebookLM available in the USGoogle opens NotebookLM, an AI notebook that answers from your own documents, to US adults.Google
  3. Gemini 1.0Google launches Gemini, a new AI that starts powering its Bard chatbot and the Pixel 8 Pro phone.Google
  4. GPT-4 Turbo preview released at OpenAI DevDayOpenAI previews GPT-4 Turbo, a cheaper GPT-4 that can read far longer documents.OpenAI
  5. Microsoft 365 Copilot generally availableMicrosoft's AI assistant for Word, Excel, Outlook and Teams goes on sale to businesses.Others
  6. DALL·E 3 in ChatGPTChatGPT can now draw pictures from your descriptions, using OpenAI's new image maker DALL·E 3.OpenAI
  7. ChatGPT can see, hear and speakChatGPT can now talk back out loud and understand pictures you show it.OpenAI
  8. Llama 2Meta releases Llama 2, a chatbot model free to download for research and for building products.Meta
  9. Claude 2Anthropic releases Claude 2 and opens a public chat website for it.Anthropic
  10. AI-powered Bing opens to all, no waitlistMicrosoft's AI-powered Bing search with built-in chat drops its waitlist and opens to anyone in preview.Others
  11. BardGoogle opens Bard, its chatbot, to people in the US and UK through a waitlist.Google
  12. ClaudeAnthropic opens its Claude chatbot beyond a closed test, to businesses that request access.Anthropic
  13. GPT-4OpenAI releases GPT-4 for paying ChatGPT users, a more reliable and creative model than GPT-3.5.OpenAI

2022

  1. ChatGPTOpenAI launches ChatGPT, a free chatbot anyone can ask questions and talk to in everyday language.OpenAI
  2. Stable Diffusion released publiclyStability AI releases an image maker anyone can download, run and use to turn text into pictures.Others