Editorial standards

Contents

1. Sources before statements
2. Licences read in full
3. Benchmarks labelled
4. Fair to every option
5. How we use AI tools
6. Corrections

Sources before statements

Every factual claim about a model, a licence, a price or a law links to a source we opened, with the access date. Where a vendor does not publish something, such as hardware guidance or a fine-tuning guide, we say so instead of estimating it.

Licences read in full

Licence terms decide what a company may do with a model, so we quote the conditions that matter (user caps, revenue thresholds, naming and branding duties, service restrictions) from the licence file itself rather than from a summary.

Benchmarks labelled

Benchmark results are published by the developers of the models. We report them as vendor-reported and do not rank on them alone.

Fair to every option

Each option is described with its strengths, its limits and the situations in which another option is the better choice.

How we use AI tools

We use AI tools to help with research, drafting and translation. Every factual claim is checked against the source it cites before publication, and translations are reviewed against the English original.

Corrections

If a fact is wrong or out of date, tell us. We fix verified errors quickly, update the page’s date and record the change in its change log.

Further reading

• Case analysisGoing headless like Revolut: what replacing Claude Code with an open model tookRevolut now runs most of its AI coding on an open-weight model it serves itself, for about a tenth of its former bill. A reconstruction of what it built, what the numbers mean for other companies, and the step that comes after.
• Practical guideHow to train your own AI model on company dataMost of the work in training a company model happens before any training starts. A step-by-step account of the data work, the choice of base model, the training methods and the rules that apply in Europe.
• Licence guideOpen source vs open weights: what each licence lets a company doAlmost every model sold as open-source AI publishes its weights but not its training data. The difference rarely matters for internal use. It matters a great deal for companies that resell a model or build a product on it.

Also in: Deutsch · Français · Español · Italiano · Nederlands · Polski