gemini-2.5-flash thinks by default and spent ~3.8k of the 4k output budget on reasoning, hitting MAX_TOKENS with a truncated JSON body -> 0 claims parsed. Set thinking_budget=0 so the full budget goes to the answer (mirrors the local path's enable_thinking=False). On the validation chunk this went from 0 -> 11 claims.
Keysat
e8d50efdf4