Key Info
Requesty reports that Jev was the most-requested model on its platform last week, yet ranked only 12th by token volume, because decision models return typed decisions rather than prose and consume roughly 12% of the tokens of an average LLM call.
Highlights
- Jev ranked #1 by requests on Requesty last week, but #12 by tokens.
- Decision models answer with a typed decision instead of an essay.
- A single call uses about 12% of the tokens of an average LLM call.
- Token-based leaderboards therefore barely reflect decision-model usage.
- Implication: request-count and token-count rankings can diverge sharply for this model class.