AI & computing, checked
Practical guides to AI assistants, API pricing, agents and running models on your own computer. Every number links to the page it came from, with the date we checked it.
Agents fit tasks that need both conversation and action, with clear success criteria and a person signing off, such as customer support and software development. Often a single model request or a fixed workflow is enough. In our worked example, model calls for a support agent handling 5,000 conversations a month cost from $17 to $2,075 depending on the model; integration, testing and review usually cost more.
Data checked
Claude or ChatGPT? Compare free and paid plans, prices, models, usage limits, context windows, coding agents and data use, sourced from both vendors.
Work out how much GPU memory a local LLM needs from its size, quantization and context length. Our formula is checked against 13 real model files.
ChatGPT, Claude, Gemini and Copilot: plans, usage limits, prices and head-to-head comparisons, with every figure linked to the vendor's own page.
2 guides
Token prices, context windows and caching for developers, plus a cost calculator for real workloads across providers.
1 guide
What AI agents are, when they are worth building, how MCP connects them to tools and data, and what automation platforms really cost.
3 guides
Run models on your own computer: how much memory they need, which GPUs and NPUs help, and the apps that make it practical.
3 guides
Tokens, context windows, retrieval and the other ideas behind today's models, explained with sources and without hype.
1 guide
The EU AI Act and the other rules, risks and security practices that apply to AI, with dates and article references.
1 guide
Seven ChatGPT alternatives compared by free plan, price, standout features and data handling: Claude, Gemini, Copilot, Perplexity, Mistral, DeepSeek, local AI.
Claude or ChatGPT? Compare free and paid plans, prices, models, usage limits, context windows, coding agents and data use, sourced from both vendors.
Ollama or LM Studio for running LLMs locally? Compare license terms, platforms, interface, local APIs and server use, from each project's own docs.
Which business tasks suit an AI agent, whether you need one, three ways to get one, what model calls cost in a worked example, the main risks and EU rules.
Install Ollama, pick a model that fits your memory, chat with it, check it runs on the GPU and connect other apps, with the official commands for each OS.
What the EU AI Act requires and from when, updated for the 2026 Digital Omnibus: prohibited practices, AI literacy, chatbot and deepfake rules, fines.
What a token is in AI, how many words it equals, why counts differ between models, how input, output and reasoning tokens are billed, and context windows.
What an AI agent is, how it differs from a chatbot or a workflow, how the plan-act-observe loop works, real examples, when you do not need one and the risks.
What the Model Context Protocol is, how hosts, clients and servers work, what tools, resources and prompts are, and the security rules the MCP spec sets.
Weights, KV cache and margin for 12 open models at any quantization and context length, checked against real model files.
Open the calculatorMonthly cost of a workload on 12 models from 4 providers, with caching, batch discounts and the tokenizer change Anthropic documents.
Open the calculatorFirst published, with a worked cost example across 12 models.
First published, with prices and features from each vendor's own page.
First published from OpenAI's and Anthropic's pricing pages and help centers.
First published, including the changes made by the Digital Omnibus on AI (Regulation (EU) 2026/1744).
First published, with a calculator for 12 open models validated against 13 real GGUF files.