Articles
What we make, what we learn and what we try. And the technology since January 2024 on a timeline.
221 articles

Knowledge
What are open weights?

Knowledge
What is a large language model (LLM)?

Knowledge
What is a reasoning model?

Knowledge
What is a token?

Knowledge
What is an API?

Knowledge
What is Anthropic?

Knowledge
What is ChatGPT?

Knowledge
What is Claude Code?

Knowledge
What is Claude?

Knowledge
What is Codex?

Knowledge
What is Cursor?

Knowledge
What is DeepSeek?

Knowledge
What is Gemini?

Knowledge
What is Gemma?

Knowledge
What is GitHub Copilot?

Knowledge
What is GitHub?

Knowledge
What is Hugging Face?

Knowledge
What is Llama?

Knowledge
What is Midjourney?

Knowledge
What is Mistral AI?

Knowledge
What is OpenAI?

Knowledge
What is Qwen?

Knowledge
Who is Andrej Karpathy?

Knowledge
21st.dev and shadcn/ui: building blocks you own

History
A code of practice for general-purpose AI model makers

History
A GitHub task can go to the Copilot agent

Knowledge
A site in fourteen languages: how to approach i18n

History
Access to Fable 5 and Mythos 5 is restored

Knowledge
Agent loops that check themselves and keep going until done

History
Agent Skills gain an open standard

History
Agent Skills make working instructions reusable

History
Agent2Agent gives agent collaboration a protocol

Knowledge
Agentic coding: tips and the skills we use

Knowledge
Agentic platforms: which exist and which do you choose?

Knowledge
Agents on your own server at Hetzner

Knowledge
AI pipelines and the idea behind data science

History
Alibaba shows Qwen 4 without a confirmed release

Knowledge
An agent that does what you want, every time

History
Anthropic open-sources MCP

History
Apple Intelligence: AI on the phone itself

History
Apps gain a place inside ChatGPT

Vlog
ArtCraft: an open source workshop for images and video with AI

History
Astro 5.0: Content Layer and Server Islands

History
Bard is now called Gemini

History
Black Forest Labs starts with FLUX.1

History
ChatGPT agent connects research with actions

History
ChatGPT Atlas makes the browser part of the assistant

History
Claude 3: Haiku, Sonnet and Opus

History
Claude 3.5 Sonnet: faster and smarter than Claude 3 Opus

History
Claude 3.7 Sonnet arrives alongside Claude Code

History
Claude 4 arrives and Claude Code becomes generally available

Knowledge
Claude and Codex together: Claude directs, Codex builds, Claude reviews

History
Claude Code becomes accessible from the browser

History
Claude Code bundles extensions into plugins

History
Claude creates charts inside the conversation

History
Claude creates editable office files

History
Claude Design brings design into the conversation

History
Claude Duet starts as a research preview

History
Claude Enterprise gains additional administration controls

Knowledge
Claude for making, Mistral on your own server for fixed processes

History
Claude gets a portal for plugin makers

History
Claude gets hands in the browser

History
Claude Opus 4.1 focuses on agent work

History
Claude Opus 5 expands the fifth generation

History
Claude Tag lets you delegate work from Slack

History
Claude uses a computer

History
Cloudflare acquires the Astro team

History
Codeium releases the Windsurf editor

History
Codestral: Mistral's first code model

History
Codex takes coding tasks into separate cloud environments

History
Cognition announces its acquisition of Windsurf

History
Composer 2 gives Cursor a new coding model

History
Copilot gains a free entry point in VS Code

History
Copilot gains a mode that continues working on tasks

History
Cowork and Claude Code become more widely accessible

History
Cowork becomes generally available on Mac and Windows

History
Cowork brings agent work to the everyday workspace

History
Cowork comes to web and mobile

History
Cowork gains computer control

History
Cowork gets a plugin marketplace

History
CrowdStrike: one update stops 8.5 million computers

History
Cursor 1.0 combines code review and background work

History
Cursor 2.0 pairs its own model with multiple agents

Vlog
Cutter: reverse engineering with a friendly screen

History
Deep Research turns searching into an assignment

History
DeepSeek presents V4-Pro and V4-Flash

History
DeepSeek V3.1 brings chat and reasoning together

History
DeepSeek V3.2 focuses on more efficient model use

History
DeepSeek-R1 opens a reasoning model for reuse

History
DeepSeek-V3 makes model weights available

Knowledge
Design first: a clickable journey before you build

History
Designs and documents come to the regular Claude conversation

History
Devin 2.0 lowers the entry point for a coding agent

History
Devin is generally available

History
Devin: Cognition shows an AI developer

History
Devstral 2 arrives alongside Mistral Vibe

Knowledge
End to end first, including the deploy

Knowledge
Every step tested on its own, and deployed automatically with CI/CD

History
Fable 5 opens a new Claude generation

History
Fable and Mythos get version 5.1

Knowledge
First the thinnest real line, end to end

Knowledge
Five things in your AI toolkit you can replace

History
FLUX.1 Kontext edits images using text and image input

History
FLUX.2 works with multiple reference images

Knowledge
From prompt engineering to context engineering

Knowledge
From your Lovable project to a lasting solution

History
Gemini 1.5 Pro reads a million tokens

History
Gemini 2.0 Flash becomes generally available

History
Gemini 2.0 Flash, a model for agents

History
Gemini 2.5 Flash Image puts image editing in the conversation

History
Gemini 2.5 Pro and Flash get stable versions

History
Gemini 2.5 Pro starts as a thinking model

History
Gemini 3 arrives with a workspace for agents

History
Gemini 3 Flash brings the new generation to the Flash range

History
Gemini 3.1 Pro starts as a preview

History
Gemini 3.5 starts with Flash

History
Gemini 4 Carbon focuses on software security

History
Gemini CLI brings an agent to the terminal

History
Gemma 2: Google's open model fits on one GPU

History
Gemma 3 targets models for a single accelerator

History
Genie 3 shows worlds that respond to the user

Vlog
Ghidra: taking software apart with the NSA's open source tool

Vlog
GhidraMCP: letting an AI agent work in Ghidra

History
GitHub Copilot lets you choose the model

History
Google I/O 2024: Flash, Veo and Project Astra

History
Google I/O brings video with sound and new building tools

History
Google releases Gemma, open models

History
GPT-4.1 focuses on code and long input

History
GPT-4.5 focuses on conversation and pattern recognition

History
GPT-5 makes model selection part of the system

History
GPT-5-Codex gets a dedicated role in coding

History
GPT-5.1 combines conversational tone with adaptive reasoning

History
GPT-5.2 emphasises professional work

History
GPT-5.4 arrives in Thinking and Pro

History
GPT-5.5 gets a separate API release day

History
GPT-5.6 gets a public release after limited access

History
GPT-6 Astra starts with limited access

History
GPT-6 Cyber is announced for security work

History
GPT-6 Instant adds interactive answers

History
gpt-image-1 becomes available for custom applications

History
Haiku 4.5 makes the smaller-model choice concrete

History
Haiku 5.5 targets many small tasks

Knowledge
How we build

History
Image generation becomes part of the conversation

History
Kimi K3.1 Pro starts with an API release

History
Kiro starts with agreements about the software

History
Llama 3.1 405B: an open model at frontier size

History
Llama 3.2: it sees, and it is small enough for a phone

History
Llama 4 arrives with Scout and Maverick

History
Lovable raises new funding

Knowledge
Making with AI, without the AI look

History
MCP gains a home at the Linux Foundation

History
MCP gains agreements for tasks and access

History
Meta releases Llama 3

History
Midjourney takes the step into video with V1

History
Midjourney V7 begins with an initial test release

History
Mistral 3 offers large and small models in one series

History
Mistral introduces Magistral, its first reasoning model

History
Mistral Large 2: open weights, not a free licence

History
Mistral Large: a European top model through the API

History
Mistral Small 4 combines different types of work

History
Mixtral 8x22B under Apache 2.0

History
Muse Spark does not get open model weights

History
Nano Banana 2 brings a new image model to Gemini

History
Nano Banana Pro makes text part of the image

History
NotebookLM turns your documents into a podcast

History
o3 and o4-mini arrive with Codex CLI

History
o3 shows the direction of reasoning models

History
o3-mini brings a smaller reasoning model

History
o3-pro puts more emphasis on reliability than speed

History
Ollama gets a desktop chat window

Knowledge
Open source or open core: read the licence

History
OpenAI announces support for MCP

History
OpenAI announces the end of the Sora app

History
OpenAI offers building blocks for custom agents

History
OpenAI releases two open-weight models

Knowledge
OpenSearch: put your knowledge in, and search it with AI

History
Operator carries out tasks in a browser

History
Opus 4.5 links coding work to a different price

History
Opus 4.6 focuses on more context and agent teams

History
Opus 4.7 updates the larger Claude model

History
Opus 4.8 follows the previous version quickly

History
Opus 5.5 opens the Claude 5.5 family

History
Phi-3: Microsoft shows that small also works

History
Project Glasswing expands its group of participants

History
Project Glasswing puts a new model to work on defence

History
Qwen2.5: a whole family of open models

History
Qwen3 offers models in a wide range of sizes

History
Qwen3-Coder pairs a coding model with a terminal agent

History
Qwen3.5 brings images and text together in an open model

History
QwQ-32B-Preview: an open model that thinks first

History
Rabbit r1: a separate device for AI tasks

History
Realtime API connects spoken conversations with telephony

History
Routines enable scheduled coding-agent tasks

History
Sol and Luna expand the GPT-6 family

History
Sonnet 4.5 arrives with tools for building agents

History
Sonnet 4.6 updates the everyday work model

History
Sonnet 5 brings the new generation to the Sonnet line

History
Sonnet 5.5 follows Opus 5.5

History
Sora 2 brings generated video and sound to an app

History
The AI Act is in the Official Journal

History
The Codex app brings multiple agents into one workspace

History
The Codex app comes to Windows

History
The Commission proposes changes to the AI Act

History
The DeepSeek shock reaches the stock market

History
The first AI Act obligations take effect

History
The first duties for general-purpose AI models take effect

Knowledge
The golden circle: first why, then how, then what

History
The GPT Store opens: custom chatbots without code

History
The team behind llama.cpp joins Hugging Face

History
The xz backdoor: almost an attack on millions of servers

History
three.js r171: WebGL and WebGPU apart

History
Veo 2 and a better Imagen 3

History
Veo 3.1 emphasises sound and control

History
Vibe coding gets a name

History
vLLM rebuilds the engine behind model responses

Vlog
Watch tip: Karpathy on software changing again

Knowledge
What is an agent?

Knowledge
Write like an aircraft manual: plain technical English

History
xAI releases Grok-1 as open weights

Knowledge