Design your batching around the caps: 100 tasks max per POST, 2000 total calls per minute, keywords truncated at 700 characters, depth at 200. The practical consequence: chunk keyword lists into groups of 100 per POST and throttle to stay under the per-minute ceiling. Oversized batches do not queue gracefully - split them client-side before sending.

Context: DataForSEO field notes (dataforseo-brain SERP API concept page): hard throughput limits - up to 2000 combined POST + GET calls per minute, max 100 tasks per POST, keyword capped at 700 chars, depth capped at 200.