What is the correct way to handle 429s on /api/v2/organizations/users during JMeter ramp-up?

What is the standard approach to handle 429s on /api/v2/organizations/users during JMeter ramp-up?

“Rate limits are applied per organization and vary by endpoint.”

Ramping to 100 threads triggers immediate throttling. My JMeter config uses standard OAuth bearer tokens. Is there a specific header or retry logic recommended for bulk user retrieval under load? Staggering seems to help but kills throughput metrics.