The living handbook of artificial intelligence · October 2026
How models work, which to use for what and what they cost, where the technology is heading. Practical guides, numbers with source and date, opinions labelled as such.
Pick the next word yourself
Learn
27 in-depth guides, each with practical examples, pitfalls and sources. Read in order or jump to what you need.
How a model really works: tokens, context, reasoning, agents. No formulas, but no wrong simplifications either.
Prompts, research, documents, automation, apps: practical methods with copy-paste examples.
Which model and plan, what it costs, what the law says, how to protect data.
Which AI for what
My picks for ten common jobs, based on Arena and Artificial Analysis rankings and daily use.
Write and edit text
Summarise long PDFs and documents
Research with sources
Excel, data and maths
Code or build a site
Create or edit images
Video and animation
Automate repetitive work
Study and learn
Work with confidential data
Models & prices
Each dot is a model: higher is more capable, further left is cheaper. The gold line links the best choice for each budget. Hover a dot.
This month in AI
Three families compete for the top: Claude (Opus and Sonnet 5.5), GPT-6 (Sol and Astra) and Gemini 4 Argon. Meta is back with Muse Wonder.
Prices fall fast: GPT-6.1 Sol completes a benchmark task for $0.72 versus almost $6 for Opus 5.5, six points apart.
Assistants are becoming software: interactive answers (ChatGPT), documents and dashboards (Claude), always-on agents.
In Europe, transparency is mandatory since 2 August 2026; high-risk rules slip to 2027-2028.
Newsletter · every Monday · free
Picked and explained by me, with one practical tip to try right away. No spam, unsubscribe in one click.
Video library
Picked from hundreds of videos watched while preparing these guides: explainers, interviews, tutorials. In Italian and English.
The perfect starting point: what an LLM is in eight minutes, with crystal-clear animations.
The four words that confuse everyone, finally sorted.
AI automates tasks, not whole jobs: the optimistic case, well argued.
The Lumen ebook
History, inner workings, tools and future of AI, explained with the lights on. For those who want to really understand, from start to finish, and keep it at hand.
Who's behind Lumen
I'm Mattia Panatta, an Economics and Finance graduate from LUISS. I use AI every day and help people, professionals and SMEs adopt it methodically: where it really saves time, under which rules, at what cost. I also build websites with an AI assistant and wrote the Lumen ebook.
Go further
Lumen explains and chooses; these sites measure, catalogue and anticipate. The chatbot checks them in real time too.
Independent benchmarks of intelligence, speed and cost for hundreds of models. The reference for numbers.
To pick an API model or see what a task really costs.
artificialanalysis.ai↗Rankings built from millions of blind, side-by-side votes by real people.
To see which model people prefer for text, code, images and video.
arena.ai↗The largest catalogue of AI tools: 50,000+ tools searchable by task.
To find the right tool for a specific, niche task.
theresanaiforthat.com↗News, previews and features in testing at OpenAI, Anthropic, Google and Meta, often before launch.
To see what's coming next. Rumours need confirming.
testingcatalog.com↗Lumen Radio
AI-generated music
Tracklist
