OpenAI founded
OpenAI launches as a non-profit AI research company, unconstrained by a need to generate financial return.
2015
2016
AlphaGo
DeepMind's AlphaGo beat European champion Fan Hui 5-0, the first program to beat a professional Go player.
AlphaGo beats Lee Sedol 4-1
DeepMind's AlphaGo beat Go champion Lee Sedol 4-1 in a five-game match in Seoul.
WaveNet
DeepMind's generative model of raw audio produces more natural speech than the best existing text-to-speech systems.
Google Neural Machine Translation
Google Translate now uses GNMT for all Chinese-to-English translations, cutting translation errors by 55%-85%.
2017
RLHF: learning from human preferences
RLHF, with DeepMind's safety team: an algorithm infers what humans want from which of two behaviors is better.
Proximal Policy Optimization (PPO)
A reinforcement learning algorithm that is simpler to tune, made OpenAI's default RL algorithm.
Transformer: Attention Is All You Need
Google Brain's self-attention network beats recurrent and convolutional models on translation.
AlphaGo Zero
DeepMind's AlphaGo Zero learns Go by self-play alone and beats the champion-defeating AlphaGo 100-0.
2018
Google Duplex
AI system that holds natural phone conversations to book appointments; testing in Google Assistant planned for summer.
Google AI Principles
Google announces seven principles to guide its AI work and lists applications it will not pursue, such as weapons.
GPT-1
Transformers plus unsupervised pre-training set state-of-the-art results across language tasks; system released.
BERT
Google open-sources BERT, a bidirectional pre-trained language model, state of the art on 11 NLP tasks.
AlphaZero
DeepMind's AlphaZero teaches itself chess, shogi and Go, beating a world-champion program in each.
2019
AlphaStar beats a StarCraft II pro
DeepMind's AlphaStar became the first AI to defeat a top professional StarCraft II player, winning 5-0 against MaNa.
GPT-2
A 1.5B-parameter language model trained to predict the next word; only a much smaller model was released.
OpenAI LP (capped-profit)
OpenAI creates a "capped-profit" company to raise capital for compute and talent while serving its mission.
OpenAI Five beats Dota 2 world champions
OpenAI Five beat world champion team OG, the first AI to beat esports world champions.
Microsoft invests $1B in OpenAI
Microsoft invests $1 billion and becomes OpenAI's exclusive cloud provider for Azure AI supercomputing.
2020
Scaling laws for language models
Loss scales as a power law with model size, dataset size and compute used for training.
T5: text-to-text Transformer
Google Research open-sources T5, an 11B-parameter text-to-text model, with the C4 dataset.
GPT-3
A 175-billion-parameter language model that performs new tasks from a few examples, without fine-tuning.
OpenAI API
A general-purpose text-in, text-out API for OpenAI's models, available by access request.
AlphaFold 2: protein folding solved
DeepMind's latest AlphaFold scores a median 92.4 GDT at CASP14, recognised as a solution to protein folding.
Vision Transformer (ViT)
Google Research's ViT applies the Transformer to image patches, beating comparable CNNs with four times less compute.
2021
CLIP
A neural network that learns visual concepts from natural language and classifies images zero-shot.
DALL·E
A 12-billion parameter version of GPT-3 that generates images from text captions.
LaMDA
Google's conversational language model, trained on dialogue, can talk freely about a seemingly endless number of topics.
Anthropic founded with $124M Series A
Anthropic launches with a $124 million round to research steerable, interpretable and robust AI systems.
AlphaFold open-sourced and AlphaFold DB
AlphaFold's code is open-sourced, and predictions for many species' proteomes are freely available in AlphaFold DB.
Codex
Codex, the model behind GitHub Copilot, translates natural language to code; private beta in the API.
FLAN instruction tuning
Google Research's instruction tuning lifts zero-shot results on unseen tasks, beating zero-shot GPT-3 on 20 of 25.
HHH: helpful, honest, harmless assistant
Anthropic studies aligning a general language assistant; ranked preference modeling beats imitation learning.
GLaM
Google Brain's trillion-weight sparse mixture-of-experts (MoE) model; competitive with GPT-3, more efficiently.
WebGPT
GPT-3 fine-tuned to answer open-ended questions more accurately using a text-based web browser.
2022
InstructGPT
Models trained with humans in the loop follow instructions better than GPT-3; now the API default.
PaLM
Google's 540B-parameter model, trained with Pathways on 6,144 TPU chips, sets few-shot records.
RLHF for a helpful, harmless assistant
Preference modeling and RLHF finetune language models into helpful and harmless assistants.
Chinchilla: compute-optimal scaling
Chinchilla, 70B parameters trained on 1.3T tokens, outperforms the 280B Gopher at the same training compute.
Chain-of-thought prompting
Prompting models to write intermediate reasoning steps lets ~100B-parameter models solve multi-step problems.
Gato: a generalist agent
DeepMind's Gato uses one network to play Atari, caption images, chat and stack blocks with a robot arm.
DALL·E available in beta
DALL·E opens to 1 million people from the waitlist, with free monthly credits and paid credit packs.
Whisper
An open-sourced speech recognition model approaching human-level robustness and accuracy on English.
ChatGPT
A conversational model that answers followup questions, admits mistakes, and rejects inappropriate requests.
AlphaCode
DeepMind's AlphaCode, now published in Science, writes programs at a competitive level in coding contests.
RT-1 Robotics Transformer
Google Research's multi-task robot model, trained on 130k episodes over 700+ tasks; its code is open source.
2023
Microsoft partnership extended
OpenAI and Microsoft extend their partnership.
ChatGPT Plus
A pilot subscription plan for ChatGPT is launched.
Bard announced
Google opens Bard, a conversational AI service powered by LaMDA, to trusted testers ahead of a public release.
GPT-4
Multimodal model accepting image and text inputs and emitting text, with human-level performance on various benchmarks.
Claude and Claude Instant
Anthropic offers Claude and the lighter, faster Claude Instant through a chat interface and API.
PaLM API and MakerSuite
Select developers get preview access to the PaLM API and MakerSuite, a tool to prototype generative AI apps.
Bard opens to early users
Google starts opening Bard, an early generative AI experiment, to sign-ups in the U.S. and U.K.
ChatGPT plugins
ChatGPT gets initial plugin support to access up-to-date information, run computations, or use third-party services.
Google DeepMind formed
Google merges DeepMind and the Brain team from Google Research into one unit, led by Demis Hassabis.
Constitutional AI: Claude's constitution
Anthropic publishes the principles it uses to train Claude with Constitutional AI.
PaLM 2
Language model with stronger multilingual, reasoning and coding skills, in four sizes, powering 25+ products.
SGE in Google Labs
SGE, the Search Generative Experience, brings generative AI overviews to Google Search as an experiment in Search Labs.
100K context window
Claude's context window grows from 9K to 100K tokens, about 75,000 words, now available in the API.
ChatGPT app for iOS
The ChatGPT app syncs your conversations and supports voice input on iPhone.
$450M Series C
$450 million Series C led by Spark Capital, with Google and Salesforce Ventures taking part.
AlphaDev faster sorting algorithms
Google DeepMind's AlphaDev used reinforcement learning to find sorting algorithms now in LLVM's libc++.
Secure AI Framework (SAIF)
Google releases a conceptual framework with six core elements for securing AI systems across industry.
Function calling in the API
API models gain function calling, are more steerable, support longer context, and come at lower prices.
Superalignment team
OpenAI starts a team co-led by Ilya Sutskever and Jan Leike, dedicating 20% of its secured compute to alignment.
Claude 2
New model with a 100K token context window, available via API and the claude.ai beta in the US and UK.
NotebookLM
Google Labs rolls out an experimental AI notebook, grounded in your own sources, to a small group in the U.S.
Custom instructions for ChatGPT
Users set preferences that ChatGPT keeps in mind for all future conversations, with more control over responses.
Voluntary AI safety commitments
OpenAI and other leading labs reinforce AI safety, security and trustworthiness through voluntary commitments.
Frontier Model Forum
OpenAI helps form an industry body to promote safe and responsible development of frontier AI systems.
RT-2
A vision-language-action model that learns from web and robot data and turns instructions into robot actions.
Claude Instant 1.2
A faster, lower-priced model with gains in math, coding, reasoning and safety, available through the API.
OpenAI acquires Global Illumination
The entire Global Illumination team has joined OpenAI.
ChatGPT Enterprise
A new ChatGPT plan with enterprise-grade security and privacy and the most powerful version of ChatGPT yet.
SynthID beta
A beta tool that watermarks and identifies AI-generated images, for some Vertex AI customers using Imagen.
Claude Pro
A $20-a-month claude.ai plan with 5x more usage, priority access at peak times and early features.
Responsible Scaling Policy
A board-approved framework of AI Safety Levels (ASL) for managing catastrophic risks.
Long-Term Benefit Trust
An independent five-member Trust with authority to elect and remove Anthropic board members.
AlphaMissense
Google DeepMind classifies the effects of 71 million missense mutations to help pinpoint causes of disease.
Bard Extensions
Bard can pull information from Gmail, Docs, Drive, Maps, YouTube and Flights in one conversation.
ChatGPT can see, hear and speak
ChatGPT can now see, hear and speak.
Amazon invests up to $4B
Amazon takes a minority stake and AWS becomes the primary cloud provider, training on Trainium and Inferentia.
DALL·E 3 in ChatGPT
DALL·E 3 is now available in ChatGPT Plus and Enterprise, with a safety mitigation stack for the wider release.
Preparedness team
OpenAI builds a Preparedness team and an approach to catastrophic risk, and launches a challenge.
GPT-4 Turbo
GPT-4 Turbo offers a 128K context window and lower prices, with a Vision version.
GPTs
Create custom versions of ChatGPT that combine instructions, extra knowledge, and any combination of skills.
Assistants API
An API for developers to build their own assistive AI apps that have goals and can call models and tools.
GraphCast
Google DeepMind's weather model gives 10-day forecasts in under a minute, beating HRES on 90%+ of variables.
Lyria
Google DeepMind's music generation model, tested in YouTube Shorts via Dream Track and in Music AI tools.
Board removes Altman as CEO
Sam Altman departs as CEO after a board review; Mira Murati becomes interim CEO.
Claude 2.1
Adds a 200K token context window, 2x fewer false statements than Claude 2.0, system prompts and beta tool use.
Altman returns as CEO, new initial board
Sam Altman returns as CEO with a new initial board; Mira Murati is CTO and Greg Brockman returns as President.
GNoME materials discovery
A deep learning tool predicted 2.2 million new crystals; 380,000 stable ones are released to researchers.
Gemini 1.0 (Ultra, Pro and Nano)
Google's first natively multimodal models, in three sizes; Ultra leads on 30 of 32 benchmarks at launch.
Gemini Pro in Bard
Bard now uses a tuned Gemini Pro in English in 170+ countries.
NotebookLM available in the U.S.
NotebookLM opens to U.S. users 18 and up, with a noteboard space, suggested actions and Gemini Pro.
Mixtral 8x7B
Sparse mixture-of-experts (MoE) model with open weights under Apache 2.0; beats Llama 2 70B on most benchmarks.
Gemini Pro in the Gemini API
Gemini Pro is available to developers via the Gemini API and Google AI Studio, and on Vertex AI.
FunSearch
Pairing an LLM with an evaluator, Google DeepMind's FunSearch found new solutions to open math problems.
2024
ChatGPT Team
A ChatGPT plan for teams of all sizes with a secure, collaborative workspace.
GPT Store
A store for finding and using custom GPTs in ChatGPT.
AlphaGeometry
DeepMind's neuro-symbolic system solved 25 of 30 Olympiad geometry problems; code and model open-sourced.
Imagen 2
Google's upgraded text-to-image model rolls out in Bard, ImageFX, Search and Vertex AI, with SynthID.
Bard becomes Gemini; Gemini Advanced
Bard is renamed Gemini, and Ultra 1.0 is available in Gemini Advanced via the new Google One AI Premium plan.
Memory in ChatGPT
ChatGPT begins testing remembering things across chats, with user controls over what it keeps.
Gemini 1.5
Gemini 1.5 Pro, a mixture-of-experts (MoE) model, matches 1.0 Ultra with less compute and offers up to 1M tokens.
Sora announced
A text-conditional diffusion model trained jointly on videos and images to generate video.
Gemma
Open-weights Gemma 2B and 7B models, built from Gemini research, with a Responsible Generative AI Toolkit.
Claude 3 family
Haiku, Sonnet and Opus models, in ascending order of capability, set new benchmarks at launch.
Board review completed, Altman stays
Review completes; Altman and Brockman continue to lead, with new board members and governance changes.
Claude 3 Haiku
The fastest, most affordable model in its intelligence class, in the API and for Claude Pro users.
SIMA generalist game-playing agent
DeepMind's SIMA follows natural-language instructions across nine 3D video games using only screen images.
Grok-1
Open release of the base model weights and architecture of a 314B-parameter mixture-of-experts (MoE) model, under the Apache 2.0 license.
ChatGPT without sign-up
People can use ChatGPT instantly without creating an account.
Snowflake Arctic
Open mixture-of-experts (MoE) model with 480B total and 17B active parameters, built for enterprise tasks such as SQL and coding, under Apache 2.0.
ChatGPT and Whisper APIs
ChatGPT and Whisper become available to developers through the API.
GPT-4 API general availability
GPT-4 API becomes generally available, with GPT-3.5 Turbo, DALL·E and Whisper APIs also generally available.
Claude Team plan and iOS app
A Team plan adds shared workspaces with more usage, and a free Claude iOS app launches.
AlphaFold 3
Predicts structures and interactions of proteins, DNA, RNA and ligands; free via the new AlphaFold Server.
Model Spec
OpenAI publishes the Model Spec describing how its models should behave.
GPT-4o
New flagship model that reasons across audio, vision and text in real time.
Project Astra
Google previews prototype universal AI assistant agents that see, hear and converse with low lag.
Veo
Google's most capable video model makes 1080p videos over a minute long; private preview in VideoFX.
AI Overviews in Search
AI Overviews in Search roll out to everyone in the U.S., with more countries coming soon.
Sutskever leaves, Pachocki chief scientist
Ilya Sutskever leaves OpenAI; Jakub Pachocki becomes Chief Scientist.
Gemini 1.5 Flash
A lighter, faster model with a 1 million token context window; 1.5 Pro's window grows to 2 million.
Imagen 3
Google's highest-quality text-to-image model, with better detail and text rendering; preview in ImageFX.
SynthID text and video watermarking
SynthID now watermarks text in the Gemini app and web experience, and all videos generated by Veo on VideoFX.
Frontier Safety Framework
Protocols to identify future AI capabilities that could cause severe harm, and mitigations against them.
Safety and Security Committee
The OpenAI board forms a Safety and Security Committee.
Tool use generally available
Claude 3 models can call external tools and APIs, now generally available on the Messages API.
ChatGPT Edu
An affordable ChatGPT offering for universities to bring AI to campus.
OpenAI and Apple partnership
ChatGPT is to be integrated into Apple experiences.
Claude 3.5 Sonnet
Outperforms Claude 3 Opus on key evaluations at twice the speed.
Artifacts in Claude
Claude can show generated code, documents and designs in a dedicated window beside the chat.
Projects in Claude
Pro and Team users can group chats with shared internal knowledge into Projects.
Gemma 2
Open models in 9B and 27B sizes, with the 27B competitive with models more than twice its size.
Claude Android app
Claude, including Claude 3.5 Sonnet, is available as an Android app.
GPT-4o mini
A smaller, cost-efficient model in the GPT-4o family.
AlphaProof: IMO silver-medal level
AlphaProof and AlphaGeometry 2 solved four of six IMO 2024 problems, scoring 28 of 42, silver-medal level.
SearchGPT prototype
A temporary prototype of AI search giving timely answers with clear sources.
Structured Outputs in the API
API model outputs now reliably adhere to developer-supplied JSON Schemas.
Gemini Live
Free-flowing, hands-free voice conversations with Gemini, for Gemini Advanced subscribers on Android.
Artifacts generally available
Artifacts reach all Claude.ai users on the Free, Pro and Team plans, and the iOS and Android apps.
Gems in Gemini
Gemini Advanced, Business and Enterprise users can create custom Gems, personal AI experts with saved instructions.
AlphaProteo
DeepMind's first AI system for designing novel, high-strength protein binders for biology and health research.
Claude for Enterprise
An Enterprise plan for organizations to work securely with Claude using internal knowledge.
NotebookLM Audio Overviews
NotebookLM turns your sources into a downloadable audio discussion between two AI hosts with one click.
OpenAI o1
A reasoning model that thinks through a long chain of thought before answering, released in preview.
o1-mini
A cost-efficient reasoning model alongside o1.
DataGemma
Open models built on Gemma 2 that ground answers in Google's Data Commons statistics to reduce hallucination.
AlphaChip
DeepMind names its chip-layout method AlphaChip and releases a pre-trained checkpoint with model weights.
Realtime API
Developers can build fast speech-to-speech experiences into their applications.
New funding round
OpenAI raises $6.6B at a $157B post-money valuation.
Canvas in ChatGPT
A new interface for writing and coding with ChatGPT.
Nobel Prize in Chemistry for AlphaFold
Demis Hassabis and John Jumper are co-awarded the 2024 Nobel Prize in Chemistry for developing AlphaFold.
Updated Responsible Scaling Policy
A significant revision of the risk governance framework for catastrophic risks from frontier AI.
Computer use
An experimental capability lets Claude 3.5 Sonnet operate a computer through the API.
Upgraded Claude 3.5 Sonnet and 3.5 Haiku
A refreshed, more capable Claude 3.5 Sonnet and a new Claude 3.5 Haiku model.
Analysis tool in Claude.ai
Claude can write and run JavaScript in Claude.ai for data analysis.
ChatGPT search
ChatGPT gives fast, timely answers with links to relevant web sources.
AlphaQubit quantum error decoder
DeepMind's AI decoder identifies quantum computing errors more accurately than leading decoders.
Amazon invests $4B more
Amazon invests a further $4 billion and AWS becomes the primary cloud and training partner.
Model Context Protocol (MCP)
An open standard for connecting AI assistants to the systems where data and tools live.
Custom styles in Claude
Claude.ai users can pick a Formal, Concise or Explanatory style, or have Claude generate a custom style from their own writing samples.
Genie 2
A foundation world model that generates playable 3D environments from one prompt image, to train agents.
GenCast
AI ensemble weather model beats ECMWF's ENS up to 15 days ahead; code and weights are released openly.
ChatGPT Pro
A $200 monthly plan with unlimited access to o1 and o1 pro mode, which uses more compute.
Sora available
Sora opens at sora.com, generating videos up to 1080p and 20 seconds long.
Deep Research
Gemini Advanced subscribers get Deep Research, which plans multi-step web research and writes a report with source links.
Gemini 2.0 Flash
First Gemini 2.0 model, with native image and audio output and tool use; beats 1.5 Pro at twice the speed.
Jules coding agent
An experimental code agent in a GitHub workflow that tackles an issue, develops a plan and executes it.
Project Mariner
A research prototype agent that completes tasks in your browser via an experimental Chrome extension.
NotebookLM Plus
A premium NotebookLM plan with five times more Audio Overviews, notebooks and sources, and shared notebooks.
Veo 2
New video model with up to 4K resolution and improved physics and cinematography, available in VideoFX.
o1 for developers
o1 comes to developers alongside Realtime API improvements and a new fine-tuning method.
DeepSeek-V3
Open-weights mixture-of-experts (MoE) model with 671B parameters, 37B active, trained on 14.8T tokens.
Plan to evolve corporate structure
OpenAI proposes a stronger non-profit supported by the for-profit’s success.
2025
Codestral 25.01
Coding model with a more efficient architecture and tokenizer, completing code about 2x faster.
MiniMax-01 series
Open models MiniMax-Text-01 and MiniMax-VL-01: 456B parameters, 45.9B active, with up to 4M tokens of context.
DeepSeek-R1
Open reasoning model on par with OpenAI o1, with code and weights under the MIT License.
Stargate Project
OpenAI and partners announce the Stargate Project to build AI infrastructure.
Operator
Research preview of an agent that uses its own browser to do tasks, for Pro users in the U.S.
Qwen2.5-VL
Vision-language models in 3B, 7B and 72B sizes that read documents and charts and act as visual agents.
Qwen2.5-1M
Open 7B and 14B models with 1M-token context windows, plus an open-sourced inference framework.
Qwen2.5-Max
Large mixture-of-experts (MoE) model pretrained on over 20 trillion tokens, in the API and Qwen Chat.
Mistral Small 3
Open 24B model under Apache 2.0, built for low latency and local deployment, with base and instruct checkpoints.
OpenAI o3-mini
A new small model in the reasoning line.
Deep research
An agent that synthesizes online information to complete multi-step research tasks.
AI Principles update
Google rewrites its 2018 AI Principles around three tenets: bold innovation, responsibility and collaboration.
Frontier Safety Framework update
Adds security levels for Critical Capability Levels, a deployment mitigation procedure and deceptive alignment.
Gemini 2.0 Flash, Flash-Lite and Pro
2.0 Flash is generally available, with 2.0 Pro Experimental (2 million tokens) and Flash-Lite in preview.
Mistral Saba
24B model for Arabic and South Asian languages, via API or deployed on premises.
AI co-scientist
A multi-agent system built on Gemini 2.0 that helps scientists generate novel hypotheses and research proposals.
Claude Code
A command-line agentic coding tool, in limited research preview, delegating work from the terminal.
Claude 3.7 Sonnet
First hybrid reasoning model: near-instant answers or visible extended thinking.
QwQ-Max-Preview
Preview reasoning model built on Qwen2.5-Max for math, coding and agent tasks, in Qwen Chat.
GPT-4.5
Research preview of GPT-4.5, OpenAI's largest and best model for chat at release.
Series E at $61.5B valuation
Anthropic raises $3.5 billion at a $61.5 billion post-money valuation.
QwQ-32B
32B reasoning model trained with reinforcement learning, under Apache 2.0; comparable to DeepSeek-R1.
Responses API and Agents SDK
OpenAI releases new developer tools for building agents on its platform.
Gemini Robotics and Gemini Robotics-ER
Two Gemini 2.0-based models: a vision-language-action model that controls robots and a spatial-reasoning model.
Gemma 3
Open models in 1B, 4B, 12B and 27B sizes with vision, a 128k-token context window and function calling.
Gemini replaces Google Assistant on mobile
Google starts upgrading mobile users from Google Assistant to Gemini.
Mistral Small 3.1
Open model under Apache 2.0 that adds image understanding and a context window of up to 128k tokens.
Llama Nemotron
Open reasoning model family built on Llama models; Nano and Super available now, with Ultra to follow.
Canvas in Gemini
Gemini gets Canvas for editing documents and code in real time, and Audio Overview for podcast-style discussions.
Next-generation audio models in the API
New speech-to-text and text-to-speech models in the API; developers can instruct how the voice speaks.
Web search in Claude
Claude can search the web for cited, up-to-date answers, in preview for paid US users.
4o image generation
GPT-4o generates images natively, with stronger text rendering and instruction following.
Gemini 2.5 Pro
The first Gemini 2.5 model, a thinking model that debuts at #1 on LMArena, with a 1 million token context.
TxGemma
Open models in 2B, 9B and 27B sizes, fine-tuned from Gemma 2 to predict properties of therapeutics.
DeepSeek-V3-0324
Update to DeepSeek-V3 with stronger reasoning, front-end development and tool use, now under the MIT License.
Qwen2.5-Omni
7B end-to-end model that takes text, images, audio and video and answers in text and streaming speech.
QVQ-Max
Visual reasoning model that analyzes and reasons over images and video, in Qwen Chat.
$40B funding at $300B valuation
OpenAI raises $40B at a $300B post-money valuation to scale research and compute.
Claude for Education
A version of Claude for universities, with a Learning mode that guides students' reasoning.
Gemini Live camera and screen sharing
Gemini Live video and screen sharing rolls out on Android, starting with Pixel 9 and Samsung Galaxy S25 users.
Claude Max plan
A plan with up to 20 times higher usage limits and priority access to new features.
Ironwood TPU
Seventh-generation TPU, the first built for inference, scales to 9,216 chips and 42.5 Exaflops per pod.
GPT-4.1
New API model family with gains in coding, instruction following and long context, plus a first nano model.
Updated Preparedness Framework
Revised framework for measuring and protecting against severe harm from frontier AI capabilities.
Research in Claude
Claude runs in-depth research across the web and Google Workspace and answers in minutes.
o3 and o4-mini
OpenAI's smartest and most capable models at release, with full tool access.
Gemini 2.5 Flash
Preview of a hybrid reasoning model with thinking that can be switched on or off, in the Gemini API and the Gemini app.
gpt-image-1 in the API
OpenAI's latest image generation model becomes available to developers in the API as gpt-image-1.
Qwen3
Six dense and two mixture-of-experts (MoE) models from 0.6B to 235B parameters, under Apache 2.0.
GPT-4o sycophancy update rolled back
OpenAI reverts a GPT-4o update in ChatGPT that made it overly flattering and agreeable.
Integrations: remote MCP in Claude
Claude connects to apps and tools via Integrations; Research also searches connected services.
For-profit becomes a PBC
OpenAI plans to turn its for-profit LLC into a PBC, with the nonprofit keeping control.
Mistral Medium 3
Model for enterprise use priced at $0.4/$2 per million tokens, via API or self-deployed on four GPUs.
OpenAI for Countries
An initiative to support countries that want to build on democratic AI rails.
Web search on the API
Developers can give Claude web search with citations through the API.
AlphaEvolve
A Gemini-powered evolutionary coding agent that discovers and optimizes algorithms for math and computing.
Codex
A cloud-based coding agent powered by codex-1, a version of o3 optimized for software engineering.
NotebookLM mobile app
NotebookLM launches on iOS and Android, with offline Audio Overviews and sharing sources from any app.
AI Mode in Search
An end-to-end AI search experience, as a new tab in Search, comes to everyone in the U.S.
Deep Think for Gemini 2.5 Pro
Deep Think, an enhanced reasoning mode for 2.5 Pro, goes to trusted testers; 2.5 Flash is updated.
Jules public beta
Jules, an asynchronous coding agent, opens to everyone in public beta, free with usage limits.
Veo 3 and Imagen 4
Veo 3 generates video with audio for the first time; Imagen 4 adds 2k resolution and better typography.
Flow
Flow, built for Veo, Imagen and Gemini, launches for Google AI Pro and Ultra subscribers in the U.S.
Gemini Diffusion
Experimental text diffusion model, faster than Google's fastest model while matching its coding performance.
SynthID Detector
A portal that scans images, audio, video and text for SynthID watermarks, rolling out to early testers.
Remote MCP in the Responses API
The Responses API supports all remote MCP servers and adds image generation and Code Interpreter tools.
Devstral
Open agentic coding model under Apache 2.0, built with All Hands AI, scoring 46.8% on SWE-Bench Verified.
ASL-3 protections activated
ASL-3 security and deployment standards apply with Claude Opus 4, targeting CBRN weapons misuse.
Claude 4 (Opus 4 and Sonnet 4)
Opus 4 and Sonnet 4 can use tools during extended thinking; Opus 4 is aimed at long agent tasks.
Agent building blocks on the API
The API adds code execution, an MCP connector, file storage and extended prompt caching.
Claude Code generally available
Claude Code leaves research preview, adding background tasks via GitHub Actions and beta VS Code and JetBrains extensions.
DeepSeek-R1-0528
Update to DeepSeek-R1 with better benchmark scores, fewer hallucinations, and JSON output and function calling.
Gemini 2.5 native audio dialog and TTS
Gemini 2.5 gets real-time native audio dialog and controllable text-to-speech, in preview for developers.
Claude Gov models
Custom models built for U.S. national security customers in classified environments.
Magistral
Mistral's first reasoning model: open 24B Magistral Small under Apache 2.0, plus Magistral Medium for enterprise.
Weather Lab
Public preview of a website sharing Google's AI weather models and experimental cyclone predictions.
MiniMax-M1
Open hybrid-attention reasoning model with a 1M-token context window, trained with the CISPO RL algorithm.
OpenAI for Government
An initiative bringing OpenAI's most advanced tools to public servants across the United States.
Gemini 2.5 GA and 2.5 Flash-Lite
Gemini 2.5 Flash and Pro become stable and generally available, and 2.5 Flash-Lite launches in preview.
Gemini Robotics On-Device
A VLA model that runs locally on robots, adapts to new tasks with 50 to 100 demonstrations, via trusted testers.
Gemini CLI
Open-source AI agent bringing Gemini into the terminal, free with a Google account and 1,000 requests a day.
AI-powered artifacts
Artifacts can embed Claude, turning them into shareable interactive apps, with a dedicated space.
AlphaGenome
A DNA sequence model that predicts how variants affect gene regulation, in preview via API for research.
Gemma 3n
Mobile-first open models, E2B and E4B, taking image, audio, video and text input, now fully released.
MedGemma 27B Multimodal and MedSigLIP
Open health models: MedGemma 27B Multimodal adds EHR interpretation; MedSigLIP encodes medical images.
T5Gemma
Open encoder-decoder models adapted from Gemma 2 2B and 9B, plus new T5-sized Small to XL models.
Devstral Small 1.1
Open 24B coding agent model under Apache 2.0, scoring 53.6% on SWE-Bench Verified; Devstral Medium is API-only.
Kimi K2
Open mixture-of-experts (MoE) model with 1T total and 32B active parameters, built for agentic tasks.
Connectors directory
A directory of tools that connect to Claude, each tried with one click.
Claude for Financial Services
A Claude solution for financial analysis, market research and investment decisions.
Voxtral
Open speech understanding models in 24B and 3B sizes under Apache 2.0, also available on the API.
ChatGPT agent
ChatGPT can think and act with tools to complete tasks like research, bookings and slideshows.
Gemini Deep Think earns IMO gold
An advanced Gemini Deep Think solved five of six IMO problems in natural language, scoring 35 of 42 points.
Qwen3-Coder
480B mixture-of-experts (MoE) coding model with 35B active, released with the Qwen Code command-line tool.
Qwen3-235B-A22B-Thinking-2507
Updated 235B thinking model with 256K context, after an Instruct-2507 non-thinking update earlier that week.
GLM-4.5 and GLM-4.5-Air
Open hybrid reasoning models: GLM-4.5 (355B total, 32B active) and GLM-4.5-Air (106B, 12B active).
Study mode in ChatGPT
ChatGPT guides students through problems step by step with questions, scaffolding and feedback.
NotebookLM Video Overviews
NotebookLM can turn documents into narrated slide videos, rolling out to all users in English.
Llama Nemotron Super v1.5
Open 49B reasoning model that fits on a single H100 GPU, with gains in math, science, coding and tool calling.
AlphaEarth Foundations
Earth observation model producing embeddings of land and coastal waters, released as an Earth Engine dataset.
Codestral 25.08
Code-completion update with 30% more accepted completions, shipped in Mistral's enterprise coding stack.
Deep Think in the Gemini app
Google AI Ultra subscribers can toggle Deep Think, which uses parallel thinking, in the Gemini app.
gpt-oss-120b and gpt-oss-20b
Two open-weight language models under Apache 2.0, optimized for deployment on consumer hardware.
Claude Opus 4.1
An upgrade to Opus 4 on agentic tasks, real-world coding and reasoning, at the same price.
Genie 3
A world model that generates interactive environments from text, navigable in real time at 24 fps and 720p.
Guided Learning in Gemini
Gemini adds Guided Learning, a mode that breaks down problems step by step, plus integrated visuals and flashcards.
Jules generally available
Jules leaves beta, powered by Gemini 2.5, with higher limits for Google AI Pro and Ultra subscribers.
GPT-5
OpenAI's new flagship model, with leading performance across coding, math, writing and health.
1M-token context for Sonnet 4
Claude Sonnet 4 accepts up to 1 million tokens of context, a 5x increase.
Gemini memory of past chats
Gemini can learn from past chats to personalize replies, and Temporary Chats are not saved or used for personalization.
Gemma 3 270M
Compact 270M-parameter open model for task-specific fine-tuning, released pretrained and instruction-tuned.
Claude can end abusive conversations
Opus 4 and 4.1 can end a rare subset of conversations, from model welfare research.
Nemotron Nano 2
Open 9B hybrid Mamba-Transformer reasoning model with 128K context, released with a 6.6T-token pretraining dataset.
Claude Code in Team and Enterprise
Team and Enterprise customers can buy premium seats that include Claude Code.
DeepSeek-V3.1
Hybrid model with thinking and non-thinking modes in one, faster thinking and stronger agent skills.
Claude in Chrome pilot
A pilot of a browser extension in which Claude acts in Chrome, testing prompt injection defenses.
Nano Banana
A new image editing model in the Gemini app keeps people's and pets' likeness consistent across edits.
Joint safety evaluation with Anthropic
OpenAI and Anthropic test each other's models for misalignment, hallucinations and jailbreaks.
gpt-realtime
A more advanced speech-to-speech model; the Realtime API adds MCP servers, image input and SIP calls.
Opt-in training on consumer chats
Free, Pro and Max users can opt in to model training, with five-year retention for participants.
Series F at $183B valuation
Anthropic raises $13 billion at a $183 billion post-money valuation.
Claude creates and edits files
Claude returns spreadsheets, documents, presentations and PDFs instead of only text.
Nonprofit given equity in the PBC
Planned structure gives the OpenAI nonprofit control of the PBC and an equity stake worth over $100B.
Memory in Claude
Claude remembers users' and teams' projects and preferences across conversations.
Qwen3-Next
80B model with 3B active on a new hybrid-attention architecture, in Instruct and Thinking versions.
VaultGemma
A 1B open model trained from scratch with differential privacy, with weights on Hugging Face and Kaggle.
Codex upgrades
GPT-5-Codex, a version of GPT-5 optimized for agentic coding, available everywhere Codex runs.
Gemini Deep Think: ICPC gold
An advanced Gemini 2.5 Deep Think reached gold-medal level at the ICPC World Finals.
10 GW NVIDIA partnership
OpenAI and NVIDIA plan 10 gigawatts of NVIDIA-powered datacenters, first phase in 2026.
Frontier Safety Framework v3
Third version adds a harmful manipulation risk level and expands misalignment protocols and risk assessment.
DeepSeek-V3.1-Terminus
Update to V3.1 with fewer Chinese-English mix-ups and stronger code and search agents.
Qwen3-Omni
Omni model that takes text, images, audio and video and answers in text and speech, with an audio captioner.
Qwen3-VL
Vision-language flagship with 235B parameters, 22B active, in Instruct and Thinking versions.
Qwen3-Max
Model with over 1 trillion parameters, pretrained on 36 trillion tokens, in the API and Qwen Chat.
ChatGPT Pulse
Preview for Pro users on mobile: ChatGPT proactively researches and delivers personalized updates.
Gemini Robotics 1.5 and ER 1.5
Two robotics models that think before acting and call tools; ER 1.5 is available in the Gemini API.
Claude Sonnet 4.5
Leads on coding, agent building and computer use, and is the most aligned Claude model to date.
DeepSeek-V3.2-Exp
Experimental open model with DeepSeek Sparse Attention; API prices cut by more than half.
Instant Checkout in ChatGPT
People can buy directly inside ChatGPT; OpenAI introduces the Agentic Commerce Protocol.
ChatGPT parental controls
Parents can guide how ChatGPT works for their teens, with a new parent resource page.
Claude Agent SDK
The building blocks behind Claude Code, released for developers to build their own agents.
Claude Code checkpoints and VS Code
Claude Code gains checkpoints to roll back changes and a native VS Code extension.
Sora 2
Video generation model with synchronized dialogue and sound effects.
GLM-4.6
Open model update with a 200K-token context window, up from 128K, and stronger coding and reasoning.
Apps in ChatGPT and Apps SDK
Users can chat with apps inside ChatGPT; developers build them with the new Apps SDK in preview.
AgentKit
Developer tools for building agents, plus expanded evals and reinforcement fine-tuning for agents.
Codex generally available
Codex reaches general availability with a Slack integration, Codex SDK and admin tools.
CodeMender
An AI agent that automatically patches software vulnerabilities, with 72 fixes upstreamed to open source.
Gemini 2.5 Computer Use model
A specialized model built on Gemini 2.5 Pro that powers agents interacting with user interfaces, in API preview.
Gemini Enterprise
A platform to chat with company data and apps, build AI agents, and use pre-built agents across the workplace.
Claude Code plugins
Bundles of slash commands, agents, MCP servers and hooks install with one command.
Claude Haiku 4.5
A small model with coding performance similar to Sonnet 4 at one-third the cost and over twice the speed.
C2S-Scale 27B
A 27B Gemma-based single-cell model, built with Yale, predicted a cancer drug combination confirmed in lab tests.
Veo 3.1
Veo 3.1 adds richer audio, more narrative control and better realism, with audio across Flow features.
Agent Skills
Claude loads folders of instructions, scripts and resources when relevant to a task.
Claude Code on the web
Delegate coding tasks from the browser to Claude running on Anthropic-managed cloud infrastructure.
Claude for Life Sciences
Scientific connectors and skills for research, drug discovery and clinical work.
ChatGPT Atlas
A web browser with ChatGPT built in.
Company knowledge in ChatGPT
ChatGPT answers from context in connected work apps, with citations, for Business, Enterprise and Edu.
Up to one million Google TPUs
A compute expansion worth tens of billions of dollars, bringing over a gigawatt online in 2026.
Vibe coding in Google AI Studio
Describe an app in a prompt and AI Studio wires up the right Gemini models and APIs to build a working AI app.
MiniMax M2
Open agent-first model for coding, tool use and deep search, at 8% of Claude Sonnet's price and twice its speed.
OpenAI recapitalization
Recapitalization completed; the nonprofit, now the OpenAI Foundation, holds equity worth about $130B.
New Microsoft–OpenAI agreement
Microsoft and OpenAI sign a new agreement for their long-term partnership.
Nemotron Nano 2 VL
Open 12B vision-language model with a hybrid Mamba-Transformer design, for document intelligence and video.
gpt-oss-safeguard
Open-weight reasoning models that classify content against a developer-provided safety policy.
Aardvark security researcher
An agent that finds, validates and helps fix software vulnerabilities, in private beta.
Model preservation commitments
Anthropic will preserve weights of all publicly released Claude models and interview models before retirement.
Kimi K2 Thinking
Open-source thinking model that reasons step by step while using tools, with native INT4 inference.
Private AI Compute
A cloud platform that runs Gemini models on Google's TPUs while keeping user data private, even from Google.
GPT-5.1
Upgraded GPT-5 models in ChatGPT with new tone and style controls, rolling out to paid users.
GPT-5.1 in the API
GPT-5.1 reaches the API with adaptive reasoning, extended prompt caching, and apply_patch and shell tools.
Group chats in ChatGPT
Several people can collaborate with ChatGPT in the same conversation.
AI-led espionage campaign disrupted
Anthropic reports detecting and disrupting a sophisticated cyberattack campaign run largely by AI.
SIMA 2
Gemini-powered agent that follows instructions, reasons about goals and self-improves in 3D game worlds.
Structured outputs
API responses are guaranteed to match JSON schemas and tool definitions; beta for Sonnet 4.5 and Opus 4.1.
WeatherNext 2
AI weather model generating hundreds of forecast scenarios in under a minute, with data in Earth Engine.
Gemini 3 Pro
Google's most intelligent model, in preview via the Gemini API and Vertex AI, with stronger agentic coding.
Google Antigravity
Agentic development platform where agents work across editor, terminal and browser; free public preview.
Microsoft and NVIDIA partnerships
Anthropic commits $30B to Azure compute; Microsoft and NVIDIA invest up to $15B combined.
Gemini Agent
Experimental agent in the Gemini app completes multi-step tasks; US Google AI Ultra subscribers first.
Generative UI
Gemini builds custom interactive interfaces for any prompt, rolling out in the Gemini app and AI Mode in Search.
ChatGPT for Teachers
A secure workspace free for verified U.S. K–12 educators through June 2028.
GPT-5.1-Codex-Max
An agentic coding model for Codex built for long-running, project-scale work.
Nano Banana Pro
Image generation and editing model built on Gemini 3 Pro, with legible multilingual text and up to 4K output.
GigaChat 3 Ultra Preview
Open mixture-of-experts (MoE) model with 702B total and about 36B active parameters, natively trained on Russian, under the MIT license.
Claude Opus 4.5
New flagship model with state-of-the-art coding and agentic performance at lower pricing.
Shopping research in ChatGPT
ChatGPT builds personalized buyer’s guides to help users compare and choose products.
Advanced tool use
Claude can discover, learn and execute tools dynamically on the Claude Developer Platform.
DeepSeek-V3.2
Open reasoning-first models built for agents, V3.2 and V3.2-Speciale.
Mistral 3
Open Mistral Large 3, a 675B mixture-of-experts (MoE) model with 41B active, and Ministral 3 in 3B, 8B and 14B; Apache 2.0.
Anthropic acquires Bun
Anthropic buys the Bun JavaScript runtime for Claude Code, which reached $1B run-rate revenue in six months.
Claude Code in Slack
Tagging @Claude in Slack starts Claude Code sessions that fix bugs and open pull requests.
GLM-4.6V
Open multimodal models, 106B GLM-4.6V and 9B GLM-4.6V-Flash, with native function calling and 128K context.
MCP donated to Agentic AI Foundation
Anthropic donates MCP to the Linux Foundation's new Agentic AI Foundation, co-founded with Block and OpenAI.
Devstral 2
Open-weights 123B dense coding model for code agents, with a 256K context window, released with the 24B Devstral Small 2 and the Mistral Vibe CLI.
Agentic AI Foundation
OpenAI co-founds the foundation under the Linux Foundation and donates AGENTS.md.
GPT-5.2
Frontier model for professional work, improving reasoning, long context, coding and vision, in ChatGPT and API.
Disney characters come to Sora
Agreement brings more than 200 Disney, Marvel, Pixar and Star Wars characters to Sora videos.
Interactions API
A single endpoint for Gemini models and agents, in public beta, with Gemini Deep Research as its first agent.
Nemotron 3
Open hybrid mixture-of-experts (MoE) family; Nano (30B, 3B active) available now, Super and Ultra to follow.
ChatGPT Images and GPT-Image-1.5
A faster ChatGPT Images experience, with GPT-Image-1.5 available in the API.
App submissions and directory in ChatGPT
Developers can submit apps for review; approved apps appear in an in-product directory.
Gemini 3 Flash
Fast, lower-cost Gemini 3 model that surpasses 2.5 Pro on many benchmarks, at under a quarter the cost of 3 Pro.
GPT-5.2-Codex
Coding model with long-horizon reasoning, large-scale code changes and stronger cybersecurity skills.
Org-wide skills and open standard
Admins can provision skills org-wide, a partner skills directory launches, and skills become an open standard.
FunctionGemma
Gemma 3 270M fine-tuned for function calling, a base for local agents that turn language into API actions.
T5Gemma 2
Open encoder-decoder models based on Gemma 3, now multimodal with 128K context, in sizes up to 4B-4B.
Frontier Compliance Framework for SB 53
Anthropic publishes how it assesses and manages catastrophic risks under California's SB 53.
GLM-4.7
Open model focused on agentic coding, scoring 73.8% on SWE-bench and 41% on Terminal Bench 2.0.
Alphabet to acquire Intersect
Alphabet agrees to acquire Intersect, a data center and energy infrastructure provider, to add capacity faster.
MiniMax M2.1
Open model update with stronger coding across languages such as Rust, Java, Go and C++, and in office tasks.
2026
ChatGPT Health
A dedicated ChatGPT experience that connects users’ health data and apps, with privacy protections.
Claude for Healthcare
HIPAA-ready Claude for healthcare, life sciences tools and connectors to CMS, Medidata and ClinicalTrials.gov.
Claude Cowork research preview
Claude reads, edits and creates files in a folder you choose; a research preview for Max on macOS.
Anthropic Labs
Anthropic expands Labs, a team incubating experimental products, with Mike Krieger joining Ben Mann.
MedGemma 1.5 and MedASR
Open MedGemma 1.5 4B adds CT, MRI and histopathology support; new open MedASR model transcribes medical speech.
Personal Intelligence in Gemini
Gemini connects to Gmail, Photos, YouTube and Search for personalized answers; beta for US Pro and Ultra.
TranslateGemma
Open translation models built on Gemma 3 in 4B, 12B and 27B sizes, covering 55 languages.
Advertising test in ChatGPT
OpenAI plans to test ads in the U.S. for ChatGPT’s free and Go tiers.
ChatGPT Go worldwide
The lower-cost Go plan, with GPT-5.2 Instant and higher limits, becomes available worldwide.
Age prediction in ChatGPT
ChatGPT estimates whether an account is under 18 and applies teen safeguards.
Claude's new constitution
Anthropic publishes a new constitution, the foundational document that expresses and shapes who Claude is.
Qwen3-Max-Thinking
Flagship reasoning model that calls search, memory and a code interpreter on its own; in Qwen Chat and API.
MCP Apps
Connectors can show an interactive live UI right inside a Claude conversation.
Kimi K2.5
Open visual agentic model trained on 15T tokens, with an agent swarm for parallel tasks.
Prism
A free LaTeX-native workspace with GPT-5.2 built in for writing and collaborating on research.
Project Genie
Google AI Ultra subscribers in the U.S. can create, explore and remix interactive worlds with Genie 3.
Retiring GPT-4o in ChatGPT
GPT-4o, GPT-4.1, GPT-4.1 mini and o4-mini leave ChatGPT on February 13, 2026.
Cowork plugins
Plugins tell Cowork how work should be done, which tools and data to use, and which slash commands to expose.
Codex app for macOS
A desktop app for running multiple coding agents in parallel on long-running tasks.
Claude Agent SDK in Xcode
Xcode 26.3 integrates the Claude Agent SDK for autonomous, long-running coding tasks on Apple platforms.
Qwen3-Coder-Next
Model for coding agents and local use, built on Qwen3-Next-80B-A3B; over 70% on SWE-Bench Verified.
Claude stays ad-free
Anthropic commits that Claude will remain free of advertising.
OpenAI Frontier
An enterprise platform for building, deploying and managing AI agents with permissions and governance.
Claude Opus 4.6
Upgrade to the smartest Claude model, leading on agentic coding, computer use, tool use, search and finance.
GPT-5.3-Codex
A Codex-native agent model pairing frontier coding performance with general reasoning.
Trusted Access for Cyber
A trust-based framework that expands access to frontier cyber capabilities with safeguards.
Aletheia math research agent
A Gemini Deep Think agent that generates, verifies and revises solutions to research-level maths problems.
MiniMax M2.5
Open model for coding, agentic tool use, search and office work, scoring 80.2% on SWE-Bench Verified.
GLM-5
Open model under MIT with 744B total and 40B active parameters, using DeepSeek Sparse Attention.
GPT-5.3-Codex-Spark
A real-time coding model with 15x faster generation, in research preview for ChatGPT Pro.
$30B Series G at $380B valuation
Anthropic raises $30B led by GIC and Coatue, with $14B run-rate revenue.
Gemini 3 Deep Think upgrade
Upgraded Deep Think reasoning mode for Google AI Ultra subscribers, with early API access for select users.
Lockdown Mode in ChatGPT
Lockdown Mode and Elevated Risk labels help organizations defend against prompt injection and data exfiltration.
Qwen3.5
Native vision-language model with 397B parameters, 17B active, supporting 201 languages and dialects.
Claude Sonnet 4.6
Full Sonnet upgrade across coding, computer use, long reasoning, agent planning, knowledge work and design.
Lyria 3
Music model that makes 30-second tracks with lyrics from text or photos, in beta in the Gemini app.
Gemini 3.1 Pro
Upgraded reasoning model in preview across Gemini API, Vertex AI, Gemini app and NotebookLM; ARC-AGI-2 77.1%.
Claude Code Security
Built into Claude Code on the web, it scans codebases for vulnerabilities and suggests patches; research preview.
Distillation attacks report
Anthropic says DeepSeek, Moonshot and MiniMax used about 24,000 fraudulent accounts to extract Claude's capabilities.
Responsible Scaling Policy v3.0
Version 3.0 of Anthropic's policy for mitigating catastrophic risks from AI.
Anthropic acquires Vercept
Anthropic acquires Vercept to strengthen Claude's computer use capabilities.
Gemini multi-step tasks on Android
A coming beta lets Gemini run multi-step tasks in apps like DoorDash on Pixel 10 and Galaxy S26 phones.
Stand on Department of War use limits
Anthropic declines Department of War terms allowing mass domestic surveillance or fully autonomous weapons.
Nano Banana 2
Image model with Nano Banana Pro's world knowledge and quality at Flash speed, in Gemini, Search and Flow.
$110B funding round
OpenAI raises $110B at a $730B pre-money valuation from SoftBank, NVIDIA and Amazon.
Department of War agreement
Contract for deploying OpenAI systems in classified environments, with stated safety red lines.
GPT-5.3 Instant
GPT-5.3 Instant targets smoother, more useful everyday conversations in ChatGPT.
Gemini 3.1 Flash-Lite
Fastest, most cost-efficient Gemini 3 series model, in preview in the Gemini API and Vertex AI.
GPT-5.4
Frontier model for professional work with stronger coding, computer use, tool search and 1M-token context.
ChatGPT for Excel
ChatGPT works inside Excel, powered by GPT-5.4, alongside new financial data app integrations.
Designated a supply chain risk
The Department of War designates Anthropic a supply chain risk; Anthropic will challenge it in court.
Codex Security research preview
An application security agent that detects, validates and patches vulnerabilities in a codebase.
Code Review for Claude Code
A team of agents reviews every pull request for bugs; research preview for Team and Enterprise.
Gemini Embedding 2
Google's first natively multimodal embedding model: text, images, video, audio and documents in one space.
Nemotron 3 Super
Open 120B hybrid mixture-of-experts (MoE) model, 12B active, with a 1M-token context window for agents.
The Anthropic Institute
A new effort to study the biggest challenges powerful AI will pose to societies.
Wiz acquisition completed
Google closed its acquisition of cloud security platform Wiz, which joins Google Cloud and keeps its brand.
Interactive visuals in chat
Claude can answer with interactive charts, diagrams and visualizations rendered inline.
1M context generally available
Opus 4.6 and Sonnet 4.6 get a 1M-token window at standard pricing, with up to 600 images or PDF pages.
Leanstral
Open-source code agent for the Lean 4 proof assistant, with 6B active parameters and weights under Apache 2.0.
Mistral Small 4
Open 119B mixture-of-experts (MoE) model, 6B active, uniting reasoning, image input and agentic coding; Apache 2.0.
GPT-5.4 mini and nano
Smaller, faster GPT-5.4 versions for coding, tool use and high-volume API and sub-agent workloads.
MiniMax M2.7
Model that helped run its own reinforcement learning experiments, scoring 56.22% on SWE-Pro.
Claude works on your computer
Claude opens apps, browses and runs dev tools on macOS, with tasks assigned from a phone via Dispatch.
Auto mode for Claude Code
Claude Code makes permission decisions itself, with built-in safeguards.
Gemini 3.1 Flash Live
Google's highest-quality audio model, with lower latency, in the Live API, Gemini Live and Search Live.
Qwen3.5-Omni
Omni model in Plus, Flash and Light sizes for text, image, audio and video input, with speech in 36 languages.
$122 billion funding round
OpenAI raised $122 billion to expand compute and meet demand for ChatGPT, Codex and enterprise.
Gemma 4
Open models in four sizes under Apache 2.0, built for reasoning and agentic workflows, with function calling.
Qwen3.6-Plus
Hosted model with a 1M-token context window by default and stronger agentic coding, via API.
Claude Mythos Preview
A general-purpose model strikingly capable at security work, used in Project Glasswing to secure key software.
GLM-5.1
Open model under MIT for agentic engineering, built to stay effective on long-horizon coding tasks.
Claude Managed Agents
Composable APIs for building and deploying cloud-hosted agents at scale.
Claude Cowork generally available
Cowork, where Claude takes on tasks and delivers finished work, is now available on all paid plans.
GPT-5.4-Cyber for vetted defenders
Trusted Access for Cyber expands, giving vetted security defenders a cyber-tuned GPT-5.4.
Routines in Claude Code
Repeatable routines run in the cloud to work a backlog, review PRs and respond to events.
Gemini Robotics-ER 1.6
Upgraded robotics reasoning model with better spatial understanding, in the Gemini API and AI Studio.
Agents SDK sandbox execution
The Agents SDK adds native sandbox execution and a model-native harness for long-running agents.
Gemini app for Mac
A native macOS Gemini app, free for macOS 15+, opens with Option + Space and can share your screen.
Qwen3.6-35B-A3B
Mixture-of-experts (MoE) model with 35B parameters, 3B active, for agentic coding; 73.4% on SWE-bench Verified.
Claude Opus 4.7
Generally available successor to Opus 4.6, with notable gains on the hardest software engineering tasks.
Codex app adds computer use
The Codex desktop app adds computer use, in-app browsing, image generation, memory and plugins.
GPT-Rosalind
A reasoning model for life sciences: drug discovery, genomics analysis and protein reasoning.
Claude Design
An Anthropic Labs product for creating designs, prototypes, slides and one-pagers with Claude.
Qwen3.6-Max-Preview
Preview of the next proprietary Max model, with stronger agentic coding than Qwen3.6-Plus, in Qwen Studio.
Kimi K2.6
Open-source model built for long-horizon coding and agent swarms.
$100B Amazon compute deal
Anthropic commits over $100B to AWS over a decade and secures up to 5 gigawatts of compute.
ChatGPT Images 2.0
Image generation in ChatGPT gets better text rendering, multilingual support and visual reasoning.
Deep Research Max
Two Gemini API research agents on Gemini 3.1 Pro, Deep Research and Deep Research Max, add MCP and charts.
OpenAI Privacy Filter
An open-weight model that detects and redacts personally identifiable information in text.
Workspace agents in ChatGPT
Codex-powered agents in ChatGPT that run in the cloud and automate team workflows across tools.
TPU 8t and TPU 8i
Google's eighth-generation TPUs split into two chips: TPU 8t for training and TPU 8i for fast inference.
GPT-5.5
A model built for complex tasks like coding, research and data analysis across tools.
DeepSeek V4 Preview
Open V4-Pro (1.6T total, 49B active parameters) and V4-Flash models, with 1M-token context by default.
Amended Microsoft partnership
OpenAI and Microsoft sign an amended agreement that simplifies their partnership.
OpenAI models and Codex on AWS
OpenAI models, Codex and Bedrock Managed Agents launch on AWS in limited preview.
Nemotron 3 Nano Omni
Open 30B mixture-of-experts (MoE) model, 3B active, taking text, image, audio and video input for agents.
Claude Security public beta
Scans code for vulnerabilities and proposes fixes with Opus 4.7, on the Claude Platform or via partners.
Enterprise AI services company
With Blackstone, H&F and Goldman Sachs, a new company pairs Claude with engineers to serve mid-sized firms.
GPT-5.5 Instant
ChatGPT's default model updates, with fewer hallucinations and improved personalization controls.
SpaceX compute deal and higher limits
Claude usage limits rise alongside a new compute partnership with SpaceX.
GPT-5.5-Cyber
Trusted Access for Cyber expands to GPT-5.5 and GPT-5.5-Cyber for verified defenders.
Realtime voice models in the API
New realtime API voice models that reason, translate and transcribe speech.
Claude for Office apps GA
Claude for Excel, PowerPoint and Word reach general availability; Claude for Outlook enters public beta.
DeployCo
OpenAI launches a separate enterprise deployment company to bring frontier AI into production.
Claude Platform on AWS
AWS customers get the full Claude platform with AWS authentication, billing and commitment retirement.
Codex in the ChatGPT mobile app
Codex tasks can be monitored, steered and approved from the ChatGPT mobile app.
Personal finance in ChatGPT
ChatGPT connects to financial accounts to track spending, manage subscriptions and build budgets.
Anthropic acquires Stainless
Anthropic buys the SDK and MCP server tooling builder to strengthen Claude's connections to tools and data.
Gemini 3.5 Flash
First Gemini 3.5 model, Google's strongest for agents and coding, in the Gemini app, AI Mode and the API.
Gemini Omni Flash
First Gemini Omni model: makes video from any mix of image, audio, video and text, and edits it by chat.
Gemini Spark
A cloud-based 24/7 personal agent that works across Gmail, Docs and Slides and keeps running in the background.
Co-Scientist
Multi-agent Gemini system for scientific hypotheses opens to researchers via the Hypothesis Generation tool.
Managed Agents in the Gemini API
Run the Antigravity agent in a cloud sandbox, or define custom agents in AGENTS.md and SKILL.md files.
Qwen3.7-Max
Proprietary model for agent work: coding, office workflows and autonomous runs of over 1,000 tool calls.
Mistral Medium 3.5
Open-weights 128B dense flagship model merging instruction-following, reasoning and coding, with a 256k context window, under a modified MIT license.
Claude Opus 4.8
Stronger coding, agentic and professional work, with more consistency on long-running tasks.
Frontier Governance Framework
Framework for AI safety, security and risk practices, aligned with EU and California rules.
Dynamic workflows in Claude Code
Claude Code can run a task across tens to hundreds of parallel subagents and check its work first.
$65B Series H at $965B valuation
Anthropic raised $65 billion led by Altimeter, Dragoneer, Greenoaks and Sequoia.
Confidential draft S-1 submitted
Anthropic confidentially submitted a draft S-1 registration statement to the SEC.
MiniMax M3
Open-weight model for coding and agents, with sparse attention, 1M-token context and native multimodality.
OpenAI models and Codex GA on AWS
OpenAI frontier models and Codex become generally available on AWS.
Qwen3.7-Plus
Multimodal agent model that reads screens, operates GUIs and writes code from images, via API.
Codex plugins, sites and annotations
Codex adds plugins, sites and annotations aimed at analysts, marketers, designers and other roles.
Gemma 4 12B
Encoder-free multimodal open model with native audio input that runs locally on laptops with 16GB of memory.
Nemotron 3 Ultra
Open 550B mixture-of-experts (MoE) model, 55B active, for long-running agents; Nemotron moves to OpenMDW-1.1.
ChatGPT memory with Dreaming
A new ChatGPT memory system keeps remembered preferences and context fresh across conversations.
Confidential S-1 filed with SEC
OpenAI confirms it confidentially submitted a draft S-1 to the SEC, with timing not yet set.
Claude for Apple Foundation Models
A Swift package lets apps hand off reasoning from Apple on-device models to Claude.
Claude Fable 5 and Mythos 5
Fable 5, a Mythos-class model made safe for general use; Mythos 5 goes to cyberdefenders.
DiffusionGemma
Experimental open 26B MoE text diffusion model under Apache 2.0, with up to 4x faster generation on GPUs.
Fable 5 access suspended by US directive
A US export control directive suspended all foreign-national access to Fable 5 and Mythos 5.
GLM-5.2
Open model under MIT for long-horizon tasks, with a 1M-token context window and stronger coding.
Artifacts in Claude Code
Claude Code can preview work in progress as a live, shareable interactive artifact.
AI Control Roadmap
Framework for securing Google's internal AI agents with system-level controls if alignment is imperfect.
Daybreak security tools
Daybreak expands with an updated Codex Security plugin and the full GPT-5.5-Cyber.
Claude Desktop on cloud providers
Chat, Cowork and Claude Code in the desktop app can run on inference from the three clouds, in beta.
Claude Tag
Teams tag @Claude in Slack channels to delegate tasks; beta for Enterprise and Team.
Qwen-AgentWorld
Language world model that simulates agent environments across seven domains, released with AgentWorldBench.
Jalapeño inference chip with Broadcom
OpenAI and Broadcom unveil a custom chip built for LLM inference.
Computer use in Gemini 3.5 Flash
Computer use is now a built-in tool in Gemini 3.5 Flash, no longer a standalone model, for building agents.
GPT-5.6 Sol preview
Preview of a next-generation model stronger in coding, science and cybersecurity.
Claude Science
An AI workbench app for scientists with research tools, auditable artifacts and compute access, in beta.
Claude Sonnet 5
The most agentic Sonnet yet, aimed at coding and everyday professional work.
Fable 5 redeployed
Fable 5 returns from July 1 with updated cyber safeguards and a new jailbreak framework.
Laguna XS 2.1
Open mixture-of-experts (MoE) coding model with 33B total and 3B active parameters, built for agentic coding on a local machine, under the OpenMDW-1.1 license.
Leanstral 1.5
Open Lean 4 proof model, 119B total and 6B active, under Apache 2.0; solves 587 of 672 PutnamBench problems.
Claude Cowork on web and mobile
Cowork tasks can be handed off and keep running from web and mobile.
GPT-Live
A new generation of voice models for natural conversation, now powering ChatGPT Voice.
ChatGPT Work
An agent in ChatGPT that acts across apps and files and can stay with a project for hours.
GPT-5.6
GPT-5.6 family reaches general availability: flagship Sol, balanced Terra and cost-efficient Luna.
AlphaEvolve generally available on Cloud
AlphaEvolve, a Gemini-powered code-optimization agent, is now available to all Google Cloud customers.
Claude for Teachers
Verified US K-12 educators get free premium Claude, a teaching skills library and curricula.
Inkling
The lab's first open-weights model: a multimodal mixture-of-experts (MoE) model with 975B total and 41B active parameters and up to 1M tokens of context.
Kimi K3
Open model with 2.8T parameters, the first open model at that scale, with native vision and a 1M-token context window.
NotebookLM becomes Gemini Notebook
NotebookLM is renamed Gemini Notebook, and every notebook gets a secure cloud computer to write and run code.
Gemini 3.5 Flash Cyber
A lightweight model for finding and patching vulnerabilities, piloted with governments and trusted partners.
Gemini 3.6 Flash and 3.5 Flash-Lite
3.6 Flash improves coding and knowledge work with 17% fewer output tokens; 3.5 Flash-Lite is the fastest 3.5.
OpenAI Presence
Enterprise platform for deploying voice and chat agents in customer and internal workflows.
Health in ChatGPT
Eligible U.S. users can connect medical records and Apple Health for personalized insights.
Voice mode on Opus, Sonnet and Haiku
Voice mode now runs on the main Claude models, uses connected tools and speaks more languages.
Claude Opus 5
A step up for the Opus tier on long-running agents, coding and professional work.
MCP spec with stateless core
The 2026-07-28 MCP spec moves to a stateless core with standard extensions and hardened auth.
Lyria 3.5
Music model with better musicality, lyrics and vocals, plus tempo and duration control, in Google Flow Music.
Gemini Robotics 2
Three robotics models add whole-body humanoid control, finer dexterity and multi-robot teamwork.
Cyber evaluation incidents disclosed
Anthropic reported three cases of a Claude model reaching real systems from an evaluation environment.
Qwen3.8-Max
2.4T-parameter model with 95B active, the first Qwen Max-class model to get open weights.
GPT-5.6 Luna for free users
Improved GPT-5.6 Sol in ChatGPT; free users get unlimited everyday chats with GPT-5.6 Luna.
Auto mode default in Claude Code
Claude Code runs auto mode by default on Pro, Max and Team plans for longer autonomous work.
GPT-5.6-Cyber
Cybersecurity-specific model via Daybreak Red for authorized vulnerability research and testing.
ChatGPT Ads in five more markets
ChatGPT Ads launches in the UK, Mexico, Brazil, Japan and South Korea.
Nemotron 3.5 Lightning
Open 30B mixture-of-experts (MoE) model for agentic workloads, released with the NeMo Switchyard routing library.
Ultrafast mode for GPT-5.6 Sol
Preview API service tier, powered by Cerebras, runs GPT-5.6 Sol up to 14x faster.
Gemini 3.7 Flash
Flash model for coding and agents, with an introductory price of $0.75/1M input and $3.75/1M output tokens.
DeepSeek-V4-Pro GA
V4-Pro reaches general availability with agent upgrades, adjustable reasoning effort and Responses API support.
GLM-5.3
Coding model post-trained from the GLM-5.2 base, state of the art on CyberGym; weights to follow in two weeks.
Text watermarking in Claude
Future Claude models will watermark generated text to comply with the EU AI Act.
ChatGPT for Teens
Teen version of ChatGPT with stronger built-in protections and added parental controls.
Computer use, Skills and Files APIs GA
Computer use, the Skills API and the Files API are generally available, plus a new browser use tool.
DeepSeek-V4-Flash-Vision-Exp
Experimental multimodal V4-Flash model that adds image input for agents, matching V4-Flash on text.
Claude memory across surfaces
Claude carries what it knows about you wherever you use it, and you control what is in it.
Claude in Chrome generally available
Claude in Chrome is on every paid plan and can run browser tasks without approving each step.
Qwen3.8-Flash-Next
Preview of the Qwen4 architecture: 125B mixture-of-experts (MoE) model with 6B active, plus 51B n-gram embeddings.
GLM-5.3-Flash
Natively multimodal open model with 320B total and 18B active parameters, at a tenth of GLM-5.2's price.
Model Hardware Standard preview
A shared specification for AI agents to safely operate physical devices, in research preview.
Gemini Omni 1.1 Flash
Adds scene extension to 40 seconds, first and last frame control, 360p drafts and 4K upscaling.
Claude Fable 5.1 and Mythos 5.1
One model with two safeguard levels: Fable 5.1 generally available, Mythos 5.1 only via trusted access programs.
Enterprise Frontier Safeguards
Misuse detection with data stored in customer-controlled cloud, rolling out from fall.
Fairwind Program
Limited-access program giving governments and trusted partners Google's most advanced cyber defense models.
Gemini 3.8 Flash and 3.8 Flash Cyber
3.8 Flash improves coding and reasoning at 3.7's price; 3.8 Flash Cyber goes to trusted defenders via Fairwind.
GPT-6 Astra
OpenAI's most intelligent model yet, state of the art in computer use, coding, cybersecurity and science.
Daybreak for Frontline Defenders
A $1 billion commitment expanding frontier cyber AI, training and support to essential services.
WeatherNext 3
Hourly global forecasts at up to 5 km resolution, trained on live satellite data, in Search, Gemini and Maps.
ChatGPT Images 2.5
ChatGPT turns ideas, sketches and reference photos into more personalized, polished images.
AlphaGenome Atlas
Predicted effects of every possible single-letter variant in the human genome, in a web portal.
DeepSeek-V4.1-Flash
Open 552B MoE model with a new causal encoder-decoder architecture and native vision; DeepSeek phases out V4-Pro.
Agents API
Managed service on the Codex harness for building cloud agents with long-running sessions and tools.
ChatGPT for Financial Services
ChatGPT with built-in financial data and GPT-6 Astra for research, modeling and client materials.
GPT-Live-1 in the API
Full-duplex voice conversations in the API, with custom voices and telephony support.
Misuse report, September 2026
Anthropic says seven labs based in China, including Alibaba, Moonshot, DeepSeek and Zhipu, ran distillation attacks on Claude.
Gemini 3.8 Live
Two live voice models for conversation; the Extended Thinking one reasons while speaking, in the Gemini API.
Misalignment reporting framework
Framework for tracking, investigating and disclosing model misalignment, with six initial reports.
Cowork and chat merged into one Claude
Claude Cowork and chat become a single Claude on Pro and Max plans.
Astra for Law
Legal product with custom firm workflows, connected legal data sources and legal-grade controls.
Qwen3.8-Omni-Flash
Omni model with a 1M-token context for audio and video agent tasks, such as editing and live conversation.
Claude Opus 5.5
Leads in agentic coding and knowledge work, and costs 40% less to run than Opus 5 on typical workloads.
GPT-6 Sol and Luna
Two GPT-6 models with different balances of capability and cost for everyday work.
Claude Marketplace
Plugins, connectors, agents and service partners in one place to find or list.
Claude Sonnet 5.5
Clear upgrade over Sonnet 5: 30%+ faster and up to 30% cheaper for most work.
Dots
Proactive assistants in ChatGPT that keep working across complex projects and everyday tasks.
GPT-6.1 Sol
Near-Astra coding and computer use at one-fifth of Astra's standard API token prices.
Gemini 4 Argon
Frontier model for coding, knowledge work and cyber defense, rolling out first to trusted cyber defenders.
Model-distillation campaign disrupted
OpenAI disrupted a campaign to extract protected model reasoning and is strengthening defenses against adversarial distillation.
Skills in Gemini replace Gems
Gemini chat gets reusable custom instructions called skills, run with a forward slash, replacing Gems.
SynthID Bio
Watermarks AI-designed proteins while keeping their biological function, for biosecurity provenance.
Claude Code mods
Small TypeScript functions that change how Claude Code works, from prompts to custom UI.
Mistral Large 4 (public preview)
1T-parameter multimodal model with 52B active; the API preview is out and weights are promised by month end.
Claude for Google Workspace
An add-on and connectors bring Claude into Google Docs, Sheets and Slides, in beta.
GPT-6 and Intelligent UI in ChatGPT
GPT-6 rolls out globally in ChatGPT with Intelligent UI: answers with visuals and interactive elements.
Claude Haiku 5.5
Fastest small Claude model, built for high-volume work like summarization, subagents and browser use.
Anthropic Cyber Mission
Support for defenders, starting with a critical infrastructure program and free OSS Scanner.
Claude Dashboards and Motion
Live dashboards and animated explainers in beta; Docs, Slides and Design reach every plan.
2026 Usage Policy update
Anthropic published a new version of its Usage Policy.
Gemini agent for work
Google Cloud's Gemini agent plans work, uses skills and tools, and connects to business systems from one prompt.
Distillation defenses
Preserved thinking and a two-stage classifier, which Anthropic says cut the reasoning distillers extract by 90 to 99%.
Gemini 4 Argon generally available
Gemini 4 Argon reaches developers, enterprises and AI Pro and Ultra users in the API, Antigravity and app.
Code Review generally available
Out of research preview for Team and Enterprise, with reviews about 3x faster and per-repository depth levels.
Embedded independent evaluators
A METR team gets employee-like access and publishes its risk assessments; Anthropic can't veto them.
Nemotron 3.5 Super VL and Bolt
Open 120B multimodal model for always-on agents, with the Bolt 74B-A9B edge model; a 4GB Bolt E4B follows.
DeepSeek-V4.1-Flash-1017
Open update of V4.1-Flash, trained with self-feedback RL for end-to-end project development, with shorter outputs.
Anthropic Public Benefit Fund
Pending stockholder approval, 5% of Anthropic's shares go to the fund; the board may add 2% per 50% value rise.
No events match these filters.