GPT‑5.4 rollout + backlash
OpenAI rolled out GPT‑5.4 across ChatGPT and introduced lighter “mini” and “nano” variants under the OpenClaw umbrella for mobile/edge inference. Legacy deep‑research mode is being retired March 26 and API access to older models (including GPT‑4o and GPT‑5 Instant/Thinking) is being phased for most users — a move that’s sparked a strong #Keep4o backlash from users who praise GPT‑4o’s personalization and crisis handling. ( )
OpenAI began rolling GPT‑5.4 mini into ChatGPT on March 18, 2026; the company says Free and Go users can access mini via the “Thinking” feature while Plus/Pro users will see it as a rate‑limit fallback and GPT‑5 Thinking mini will be removed as a selectable option after 30 days. (help.openai.com) OpenAI describes GPT‑5.4 mini and nano as smaller, faster variants tuned for coding, multimodal tool use and high‑volume sub‑agent workloads, and says mini approaches full GPT‑5.4 performance on several internal and third‑party evaluations. (openai.com) Developer posts and OpenAI community notes list GPT‑5.4 mini with a ~400k‑token context window and show API pricing examples of roughly $0.75/$4.50 per 1M tokens (input/output) for mini and about $0.20/$1.25 per 1M tokens for nano. (community.openai.com) OpenAI set a firm removal date for the legacy deep‑research mode of March 26, 2026 while explicitly saying the current deep‑research experience will remain available and that historical conversations and results will stay accessible. (help.openai.com) OpenAI retired GPT‑4o and several older models from ChatGPT on February 13, 2026, but told developers that API access “remains unchanged” for now and allowed ChatGPT Business/Enterprise/Edu workspaces to retain GPT‑4o in Custom GPTs only through April 3, 2026. (help.openai.com) The removals sparked a #Keep4o movement across X and Reddit and prompted reporting that users prize GPT‑4o for its conversational warmth and personalization, with commentators warning the cut exposed how quickly familiar AI companions can be taken away. (piunikaweb.com) OpenAI’s own blog notes the company previously deprecated then restored GPT‑4o after user feedback—saying that feedback showed some customers relied on GPT‑4o for creative ideation and a preferred conversational style. (openai.com)
