Tests show ChatGPT Instant mode silently drops or truncates responses

A heavy ChatGPT user's A/B/A tests on GPT-5.6 and GPT-6 showed Instant mode repeatedly dropped or truncated responses while Medium mode succeeded, prompting public criticism that paying users are being treated as free QA engineers.

2026-10-13 ~ 2026-10-13 · 2 related posts

• A/B/A tests show GPT-5.6 and GPT-6 both fail on Instant while Medium works, with HTTP 200 but vanishing replies — echoabenteuer · 2026-10-13
• "Paying users shouldn't be unpaid QA": power user publicly challenges OpenAI over Instant reliability — echoabenteuer · 2026-10-13