ChatGPT Plus and Pro stop switching to Thinking for you, and GPT-5.5 retires October 14
ChatGPT Plus and Pro subscribers now have to ask for deeper reasoning themselves, and GPT-5.5 is on its way out of ChatGPT. OpenAI's release notes say it retired automatic switching from Instant to Thinking for Plus and Pro users globally on September 14, and a changelog entry the same day sets October 14 for GPT-5.5's retirement from ChatGPT, ChatGPT Work and Codex.
Asking ChatGPT to think harder no longer works
The change also removes the Higher intelligence setting from ChatGPT on the web for both plans. You can still choose a thinking level in the model picker, and ChatGPT can still switch automatically for safety purposes.
OpenAI's help article on GPT-5.6 spells out what that means in practice: "Asking ChatGPT to 'think harder' or 'think deeply' won't automatically switch to a different thinking level." OpenAI gives no reason for the change.
The Thinking slider runs from Instant through Medium, High and Extra High, all powered by GPT-5.6 Sol, to a Pro level that uses the Pro models. Plus gets Medium and High only, while Pro, Business and Enterprise get every level.
Business keeps its Higher intelligence setting under Settings > General. Free and Go users get a Think button instead, which runs the smaller GPT-5.6 Luna.
A slide made by GPT-5.5 for OpenAI's GPT-5.6 launch post. The older model leaves ChatGPT on October 14. Image: OpenAI
GPT-6 Pro comes with tight message caps
The same article lists how much GPT-6 Pro each plan gets in Chat. GPT-6 Pro runs on GPT-6 Astra, OpenAI's newest model, introduced on September 3:
- Pro $200: 200 messages per week, plus a separate daily allowance for GPT-5.6 Sol Pro
- Pro $100: 50 messages per week, shared with GPT-5.6 Sol Pro
- Business Standard: 15 messages per month, shared
- Business Premium: 50 messages per week, shared
When Pro $200 users reach the weekly limit, ChatGPT "automatically switches to GPT-5.6 Thinking at Medium". Getting onto that tier is harder for now. As of September 18, OpenAI's Pro tiers page says new sign-ups and upgrades to Pro $200 have been temporarily paused since September 10. Existing Pro $200 subscriptions renew as usual, and Pro $100 is not affected.
GPT-5.5 leaves on October 14, but not the API
The changelog says GPT-5.5 will retire "on all plans, including consumer, Business, Enterprise, and Edu plans", adding that "this retirement does not apply to the OpenAI API." Codex users who sign in with ChatGPT are told to move from gpt-5.5 to gpt-5.6-sol and to update saved model settings, custom agents, scheduled tasks and scripts that still select the old model.
One Codex model has gone already. GPT-5.3-Codex-Spark was deprecated on September 14, and the research preview is no longer available in the ChatGPT desktop app, Codex CLI or IDE extension.
OpenAI has cleared out models all summer. GPT-4.5 left ChatGPT on June 26, with its conversations handed to GPT-5.5, and nine days later GPT-5.6's public launch looked days away. Now GPT-5.5 gets the same treatment.
Anyone with a Codex script pinned to gpt-5.5 has until October 14 to change one line.
