Understanding AI models

The foundations: what AI models are, the popular ones, how to choose between them, what they are used for, and where the field stands today.

01
What is an AI model
Parameters, training and inference, tokens, and how a language model produces text.
Coming soon
02
Types of models
Language, embedding, reranking, vision, speech, multimodal and image generation models, and classic ML.
Coming soon
03
Popular models
The main closed and open-weight model families, and what each is known for.
Coming soon
04
Open-weight vs closed models
Licences, hosting, data sovereignty and cost: when each makes sense.
Coming soon
05
Model sizes and specs
Parameters, context window, quantisation and GPU memory, and how to read a model card.
Coming soon
06
Benchmarks and leaderboards
What the common benchmark scores measure, and why to test on your own data too.
Coming soon
07
How to choose a model
A checklist: task, quality, speed, cost, hardware, language support, licence and data location.
Coming soon
08
Prompting, RAG and fine-tuning
The ways to adapt a model to your organisation, and when to use each.
Coming soon
09
Use cases
Proven uses in banking, government, defence, telecom and healthcare, with examples.
Coming soon
10
Running models
Cloud APIs versus self-hosting, inference servers, and sizing GPUs.
Coming soon
11
The current landscape
Where AI stands today: reasoning models, agents, multimodal and small models, and sovereign AI.
Coming soon
12
Risks and responsible use
Hallucination, bias, security and privacy, and how to govern AI use.
Coming soon