PAGES32
WORDS50,403
SOURCES235
READ TIME94 min
TYPEHUB
CHECKED25 AUG 26
One of the larger topics on the site. What these systems are, what they can be shown to do, and how to choose between them.
What they are
The mechanism, in plain terms, before the benchmarks.
EXPLAINER · 12 MIN · 20 SOURCESAI Terms People Get Wrong: Open Source, Reasoning, Memory and MoreOpen source, reasoning, hallucination, token, memory, AGI and more: the common wrong use of each AI term, what it actually means, and why it matters.EXPLAINER · 4 MIN · 4 SOURCESAstra, Opus, Flash: What AI Model Names Actually MeanFour vendors, four naming systems, none of which tells you the price. What each scheme encodes, where a higher number misleads, and the rule that survives.COMPARISON · 3 MIN · 3 SOURCESWhich OpenAI Model to Use: The Four That MatterThe catalogue lists over a hundred models. Four are current. What each is for, how to read the rest, and the word in a model name that should stop you.COMPARISON · 4 MIN · 5 SOURCESWhich Claude Model to Use, and What Each CostsAnthropic names its own default. The four current models with prices, context windows, knowledge cutoffs — and the one that could retire first.EXPLAINER · 4 MIN · 4 SOURCESSAM 3.1: What Meta Changed, and Why It Is Twice as FastMeta’s segmentation model now tracks up to sixteen objects in one pass instead of one pass each. The published figures, and where the model already runs.COMPARISON · 3 MIN · 2 SOURCESGemma 4 vs Muse Glimmer: Two Open Models, Two Different BetsGoogle shipped four sizes in April; Meta shipped one in August. Both Apache 2.0. What each was built for, and which questions neither vendor answers.COMPARISON · 3 MIN · 3 SOURCESGrok 4.6 vs Grok 4.5: What Actually ChangedFour weeks apart, same price. The published benchmark gaps, the one behaviour change worth noting, and why Grok 4.3 is not in this comparison.EXPLAINER · 3 MIN · 1 SOURCESGemini Notebook, Formerly NotebookLM: What Actually ChangedGoogle renamed NotebookLM on 16 July 2026 and added a secure cloud computer, so notebooks can run code. The old link still redirects.EXPLAINER · 4 MIN · 5 SOURCESGemini Temperature Setting Deprecated: What ChangedGoogle deprecated temperature, top_p and top_k in the Gemini API on 21 July 2026 and tells developers to strip them. What replaced them, and what still works.EXPLAINER · 1 MIN · 1 SOURCESHOW AN AI IS BUILTFrom sand to answer: silicon, data, pretraining, alignment, products, agents.EXPLAINER · 4 MINUnderstanding AI Model Types: Image, Video, Audio & LLM ExplainedWhat each type of AI model does, what it cannot do, and how image, video, audio, and LLM models work together in a production pipeEXPLAINER · 3 MINThe Eight Kinds of AI You Already UseEight entities live in the machine city: the Oracle-Poet, the Librarian, the Swift, the Artist, the Commune, the Titan,…EXPLAINER · 7 MINThe AI Glossary: sixty terms, honestly definedSixty AI terms defined in plain language as used in mid-2026: tokens, transformers, RAG, quantization, MoE, model…
What they measurably do
Sourced and dated. Re-queryable where possible.
EXPLAINER · 4 MINHOW AI IS TESTEDBenchmarks, red teams, interpretability, monitoring — what each catches, what each misses, and the nine questions none…EXPLAINER · 3 MIN · 3 SOURCESWHY AI MAKES THINGS UPWhy AI models invent facts: scoring rewards a lucky guess over “I don’t know”, and rare facts cannot be recalled. Four mechanisms, and what actually reduces it.THE RECORD · 4 MIN · 8 SOURCESHow AI Influences You BackEveryone writes about how prompts shape models. The other direction is measured, substantial and largely invisible: sycophancy, foEXPLAINER · 3 MIN · 2 SOURCESDOES AI UNDERSTAND?Both confident answers are wrong. Explained plainly, with sources named and dated.
Choosing between them
Current status, including the Sora API, which OpenAI scheduled to shut down on 24 September 2026.
DATED · 3 MIN · 3 SOURCESThe Sora API Shuts Down on 24 September 2026OpenAI is removing the Videos API and every Sora 2 model, and its own table names no replacement. What goes, what breaks, and where the alternatives are.THE CRAFT · 18 MIN · 6 SOURCESModel GuidesAround twenty models and studios with their optimal prompt format — ChatGPT Images 2.5, Seedance, Kling, FLUX, Midjourney, Suno, ElTHE CRAFT · 4 MIN · 1 SOURCESBad, Good, GreatThe same goal written three ways. The fastest way to see what actually separates an amateur prompt from a production one.THE CRAFT · 2 MINPrompt, Context, Skill, or Fine-TuneFour ways to shape model behaviour at very different costs. Most fine-tuning problems are context problems, and a fine-tune is lost when its base model retires.THE CRAFT · 3 MIN · 1 SOURCESWhat You Are Actually BurningYou pay for everything the model reads, re-sent every turn. Where the cost really goes, why agents multiply it, and the model choice that fixes most of it.EXPLAINER · 3 MINThe AI Tools Atlas: the landscape by job, not by logoThe AI tool landscape mapped by what you are trying to do: assistants, coding agents, image, video, voice, answer…
Working with them
The craft that survives a model release.
THE CRAFT · 3 MIN · 2 SOURCESContext EngineeringWhat surrounds a prompt — the material, the history, the tools, the memory — now matters more than its wording. Four rules that hold as the tooling changes.EXPLAINER · 4 MINThe CARE Framework: Context, Anchor, Restrict, ExemplifyA four-part AI prompting framework that works first time. CARE — Context, Anchor, Restrict, Exemplify — applied to image, video, aEXPLAINER · 3 MINWhat Makes a Prompt Good? The 6 Non-Negotiable ElementsSix elements that separate AI prompts that work from ones that don't. Applies to image, video, and LLM models. Learn what every suTHE RECORD · 4 MINHOW TO ASK AI WELLNot prompt tricks that expire with the next release. Sourced, dated, and revised when the numbers move.EXPLAINER · 4 MINPrompting for LLMs: ChatGPT, Claude & Gemini — What Actually DiffersPractical differences between prompting ChatGPT, Claude, and Gemini. Model-specific strategies, system prompt structures, and how
Where they run
Sand, power and heat.
EXPLAINER · 6 MIN · 1 SOURCESWhat AI Actually Runs On: sand, power, and heatDescend from silicon die to power grid: why GPUs, what memory bandwidth means, how a training run works, and the…EXPLAINER · 5 MIN · 2 SOURCESLocal or Cloud: Where Does AI Live Next?A no-sides study of local versus cloud AI in mid-2026: the roughly 3-6 month capability gap, the strongest honest case…EXPLAINER · 3 MIN · 2 SOURCESOPEN OR CLOSEDOpen weights trail the frontier by months, not years. Explained plainly, with sources named and dated.DATED · 4 MIN · 7 SOURCESENERGY AND WATEREstimates of the water behind one AI query differ a thousandfold. Why they disagree, what operators disclose, and why where the water comes from matters most.