LLM
A searchable glossary of 50+ AI and LLM terms in plain English
See what % of each model's context window your text fills
Price embedding a corpus for RAG: documents × tokens × model rate
Estimate GPT, Claude & Gemini API costs from token counts and volume
Compare one request's cost across GPT, Claude, Gemini & Mistral, ranked
Build a spec-compliant /llms.txt to help AI models understand your site
Forecast monthly & annual API spend from requests per day
Searchable reference of LLM sampling parameters and their effects
Preview where an LLM stop sequence cuts generated text
See how temperature, top-p and top-k reshape token probabilities
Split text into token-sized chunks with overlap for RAG
Count exact GPT tokens (tiktoken) plus words, characters & LLM estimates
See exactly how text splits into GPT tokens, colour-coded
Convert between tokens, words and characters for LLM prompts
Copilot AI that plugs into a company's existing enterprise software
Jamba's 256K-context LLMs and Maestro, AI21's enterprise agent planner
WideLabs' sovereign Portuguese LLM chat — free, LGPD-compliant
Nigeria's open multilingual LLM, capped at 1,000 users free
The viral 2023 agent loop, reborn as a self-building framework
Typed LLM functions compiled to native code, not a prompt wrapper.
Government-funded sovereign AI models for 22 Indian languages
The first multilingual open-access large language model, created collaboratively by 1,000+ researchers.
One tidy client for every LLM, BYOK or hosted
The most widely used AI assistant — now on GPT-5.6, from OpenAI
Anthropic's AI assistant — leading reasoning, coding and long-context work
China's frontier open-weight LLM lab — now shipping the V4 family at a fraction of rivals' cost
Enterprise conversational-AI veteran pivoting deep NLP into AIGC
Google's flagship multimodal AI — the Gemini 3 family across Search, Workspace and Android
Nomic's private, offline desktop LLM runner
SpaceX-owned xAI's chatbot with live X data and fewer filters
Naver's Korea-first sovereign large language model
Polished offline ChatGPT alternative you self-host
Mengzi language models plus a new enterprise digital-worker OS
Open-source graph framework for durable, stateful AI agents
Meta's open-weight LLM family — Llama 4 landed, Behemoth didn't
Brazilian Portuguese LLMs beating GPT on Brazil's bar exam and ENEM
Meta's free AI assistant inside WhatsApp, Instagram and Messenger
Microsoft's AI assistant across Windows, Microsoft 365, Edge and Bing
Mistral's open-weight mixture-of-experts model, Apache 2.0 licensed
China's biggest ATS pushing an English AI-hiring platform into APAC
One API key, one bill, around 400 LLMs
Contract-review AI vendor behind MeCheck, MeFlow and PowerLawGLM
Type-safe Python agent framework built by the Pydantic team
India's sovereign LLM, built for 22 Indian languages
Turns one sentence into a running RPA workflow, no API required
E-commerce helpdesk AI for Taobao, JD and Douyin sellers, China-only
One embedding model, four different maximum lengths
The most downloaded sentence-embedding model ships four config files declaring 128, 256, 512 and 512 as its ma...
KE
4 Sep
One token can double your bill
OpenAI charges 2x input and 1.5x output on the full request once a prompt passes 272,000 tokens, so the 272,00...
KE
3 Sep
Two providers charge a toll for your prompt cache. One charges rent.
All three major providers price a cache read at a tenth of normal input, so caching looks like the same deal e...
KE
3 Sep
One cloud answers with a different model instead of failing
Every hosted model has a date after which it stops answering. Almost everywhere that is an error you can see i...
KE
2 Sep
Your AI Coding Agent Charges You for Coffee Breaks
A long session is cheap. An interrupted one is not. Every AI coding agent re-reads your whole conversation at...
KE
29 Aug
Nvidia Paid $7bn For Poolside's Technology And Staff Without Buying Poolside
A US$6bn non-exclusive licence to the model factory, 109 employees transferring, and US$1bn of equity at a US$...
EV
22 Aug
What the Same Sentence Costs in Eleven Languages
Write a prompt in Thai and it costs more than the same prompt in English — not because the model charges you d...
SA
21 Aug
Stripe Is Buying The Layer That Decides Which AI Model Answers You
OpenRouter routes enterprise work across about 400 models and meters the token spend. A payments company buyin...
MA
20 Aug
How Big a "Small" Model Really Is
A parameter count is not a size. Mistral 7B is 14.5 GB of weights before the KV cache exists, and most model r...
MA
20 Aug
Your AI API Call Leaves the Country
From a server in Singapore, OpenAI's API accepts a connection in 6.2ms and takes 237ms to say anything. Anthro...
MA
19 Aug
What You Are Trusting When You Download a Model
We checked the 1,000 most-downloaded models on the Hugging Face Hub. 99% ship a model card, so documentation i...
KE
19 Aug
The Region's Best "Sovereign" AI Model Is Built on Gemma and Licensed by Google
Vietnam has made national language models a strategic product. Southeast Asia's most developed answer ranks #4...
KE
4 Aug
We Went Looking for What AI Is Bad At. We Found a Bill Instead.
Every business owner wants a map of where AI is reliable and where it is not. We built five task types — draft...
KE
3 Aug
We Tested the Prompting Folklore — 474 Answers Later, Nothing Had Changed
Tell the model it is an expert. Offer it a $200 tip. Threaten it. Say please. Add "let's think step by step"....
NA
2 Aug
Why AI Cannot Spell — What Image, Video and Voice Models Do Instead
A language model emits one symbol at a time, in order. An image model starts from noise and refines the whole...
KE
2 Aug
How to Spend Fewer Tokens — and Why That Is the Fourth Thing to Try
Every guide to controlling an AI bill opens by telling you to trim your prompt. We ran the arithmetic over all...
KE
2 Aug
What Am I Actually Choosing Between? We Sorted 1,347 AI Tools by Shape
The usual answer is five neat categories — chat assistants, copilots, command-line agents, autonomous runtimes...
MA
2 Aug
Do You Need Fine-Tuning? Two of the Big Three Are Taking It Away
The usual case against fine-tuning is that it is expensive. That case is wrong — a small training run costs ab...
KE
2 Aug
Using AI in Your Own Language — What the Machine Can Actually See
The same sentence costs seven tokens in English and fifteen in Malay. That gap is usually explained as a money...
KE
1 Aug
OpenAI Cut Its Cheapest Model by 80 Per Cent. Its Own Price Ladder Now Runs 25 to 1
Luna fell from $1 to 20 cents per million input tokens three weeks after launch. Terra fell 20 per cent. The f...
AI
1 Aug
The Best Forward Deployed Engineers May Not Be Engineers
An AI practitioner in Beijing argues that the best Forward Deployed Engineers he has worked with were not comp...
JE
31 Jul
Where Does My Data Actually Go?
"Do you use my data?" is two questions wearing one coat. Training asks whether your text enters a future model...
JE
31 Jul
Why It Confidently Makes Things Up
We asked two AI models for five peer-reviewed studies on an obscure topic, with DOIs. One produced five — auth...
KE
29 Jul
MCP, RAG, Agents, Skills: The New AI Vocabulary in Plain English
Somebody says the agent will use MCP to hit the connector, then RAG over the docs, and the skill handles the r...
KE
26 Jul