Key Info
Michael Friedmann of Anwalt.de reports running 77B tokens of legal AI workloads through Requesty — 66.9B input and 7.0B output tokens — while keeping control over data routing, model assignment, and spending in one place.
Highlights
- Total volume: 77B tokens processed via Requesty
- Breakdown: 66.9B input tokens and 7.0B output tokens
- Workload is legal AI, a domain with strict data-handling requirements
- Emphasis on centralized control: where data goes, which models run which tasks, and cost
- Illustrates a large-scale production deployment of an AI gateway
