ChatGPT's web retrieval internals leak via server-sent events debug stream
metehan777 · x · 2026-09-08
Blogger Metehan found detailed debug data inside ChatGPT's server-sent events stream, exposing its web retrieval system: which queries it writes, which search engines it calls, returned results, scoring objects, page fetching, chunking strategy, and what makes it into the final answer. He has been pulling this data for days and published analysis on Substack—an unofficial leak of ChatGPT's retrieval pipeline, directly useful for GEO/SEO work.
Related event: Dev uncovers ChatGPT's full search debug data in SSE stream(2 posts)→
More from Models
- MissionReport benchmark: only 3 of 540 AI-generated government reports pass all criteria — Exp_Mark · 2026-10-27
- Local beats cloud: Qwen 3.8 FN handles private student data as well as frontier models — dr_lm · 2026-10-27
- Aidan Gomez: model's CoT said no tool calls right before making them — aidangomez · 2026-10-27
- Nat Lambert: Cohere's model is narrow RL training on internal infra ill-suited for open weights — natolambert · 2026-10-27
- OpenAI teases DOTS is coming to Europe 'soon' — mark_k · 2026-10-27
- Clifford Pickover challenges AI to decipher a message he encoded in Norse runes — pickover · 2026-10-27