perf: raise rate limit to 200/min for paid-tier models

feat: backend