Intermittent Multiple Responses in Single Output with gpt-5.2-chat-latest

Cross-linking — same behavior on gpt-5.5 (thinking) at the raw API level (Responses API), still present months later. A cross-model A/B isolates it to gpt-5.5 (gpt-4.1 0/30, claude-sonnet-4-6 0/30), and token billing confirms the model itself generates the duplicate — clean openings bill ~77–93 output tokens, doubled ones ~130–258. Consolidated evidence: Regression: Double responses in apps

Happy to share request IDs / a sanitized payload.