OpenAI adds GPT-5 Auto, Fast, and Thinking settings to ChatGPT's model picker; Auto seems to work like GPT-5's model router that OpenAI initially announced
When OpenAI launched GPT-5 last week, the company said the model would simplify the ChatGPT experience.
Context & Ripple Effects
OpenAI had framed GPT-5 as a unified system: an efficient model for routine requests, a reasoning model for harder tasks, and a router that selects between them. The new picker makes that architecture visible to ChatGPT users rather than leaving all model selection implicit.
The change also sits alongside OpenAI's earlier plan to differentiate GPT-5 access and intelligence levels across free and paid ChatGPT tiers, as outlined in its GPT-5 access plan. It matters because the product is now balancing a simplified default with explicit controls for users who want to trade speed against deeper reasoning.
First-order effects
- ChatGPT users gain three GPT-5 choices—Auto, Fast, and Thinking—so they can either defer selection to Auto or directly choose a speed- or reasoning-oriented mode.
- If Auto operates as the announced router, OpenAI can retain a single default experience while directing individual prompts toward different GPT-5 capabilities.
Second-order effects
- The picker creates a clearer user-facing distinction between quick responses and deliberative work, increasing pressure on assistant products to offer both low-friction defaults and intelligible control over model behavior.
- The labels make routing quality more consequential: users can compare Auto with Fast or Thinking directly, making mismatches between task and selected mode more visible.
Third-order effects
- This points toward assistants being sold as adaptive systems rather than as one fixed model: automatic routing becomes the default layer, with manual overrides for users who value predictability.
- If this pattern persists, model choice may shift from brand-level comparisons to product-level orchestration—how well an assistant allocates capability, latency, and user control across tasks.
The trend: Consumer AI assistants are evolving from single-model chatbots into routed, multi-mode products that hide complexity by default but expose controls when users need them.