01
Where this one gets misread
Tokens are not words, so neither a cost estimate nor a context limit can be worked out from a word count. The same text tokenises differently across models, which makes vendor comparisons on this basis unreliable.
Glossary
What is a token?
A token is a unit of text or other input as a model processes it. Tokens are not words, so a limit or a cost cannot be worked out from a word count alone.
Updated 21 Aug 2026
02
Questions to ask
Ask for a worked example on your own material, and the evidence needed to reproduce it.
- How does your own corpus tokenise?
- Is billing counted on input, on output, or on both?
- Where are the limits, and what happens at them?
- How does all of this change on a model swap?
03
How Marella uses the term
We use “Token” only where a product mechanism or an evaluation method backs it up, and we say when the behaviour depends on how a deployment is configured.
- Backed by a product mechanism or an evaluation method
- Deployment differences flagged
What this page does not prove
- B1A definition is not a claim about how the product performs.
- B2Vendor implementations vary.
- B3Test the term against a representative workflow.
Related pages
Test the claim on your documents
Pick a real piece of work, agree what a good answer looks like, then go through the results together.