OpenAI reverses its own policy after backlash over GPT‑5 launch
What unfolded?
On August 6, OpenAI CEO Sam Altman announced the debut of GPT‑5, a model billed as the unified successor to older versions. Immediately after the livestream, the firm disabled public access to legacy models, forcing every user to interact exclusively with the new iteration.
The move sparked a wave of fury across Reddit and other forums. Users had spent months crafting prompts tailored to GPT‑4o, the “personality” that felt like a friend; removing the option felt like erasing a personal relationship.
Why did the company pull the plug?
OpenAI’s logic was simplicity. By eliminating older models, the platform could focus resources on GPT‑5, which promises to combine the best elements of previous iterations into a single, streamlined experience.
Immediate consequences for users
- A small group of early adopters gained temporary, limited access to legacy models.
- Reddit users reacted with exasperation; one even described vomiting at the loss of GPT‑4o’s “personality.”
- Altman took part in a Reddit live Q&A, where many voiced disappointment at GPT‑5’s sanitized character and questioned whether the new model truly reflected users’ expectations.
OpenAI’s pivot: reopening legacy options
Within days, the company announced that ChatGPT Plus subscribers—those paying $20/month—could again choose from older models. However, the service currently offers limited options, namely GPT‑4o and GPT‑3.
Meta‑level takeaway: When a new, ambitious AI model is rolled out, retaining access to earlier versions can be crucial for user trust. OpenAI’s decision to let users fall back on familiar models demonstrates responsiveness to community sentiment, even after an initial hard‑edge policy shift.

OpenAI GPT‑5 Release: Subscriber Sentiment and Model Evolution
Subscriber Response – OpenAI witnessed a notable drop in its user base following the rollout of GPT‑5. The brand has responded by hosting a Reddit AMA and incorporating direct feedback into subsequent model tweaks, signaling a priority to retain existing subscribers.
Positive Highlights of GPT‑5
- Enhanced parallel task handling and advanced coding capabilities.
- Reduced hallucination incidents and a more professional communication tone.
- “Safe completions” reinforce a balance between avoiding dangerous requests and providing genuine assistance.
Critical Comparisons
Writing Capability – While GPT‑5 showcases improvements across many functional areas, its prose output is still considered inferior when measured against GPT‑4o and GPT‑4.5.
Altman’s Strategic Perspective
CEO Sam Altman admitted that the model’s launch under‑estimated the importance of features that GPT‑4o users truly value, even though GPT‑5 performs better overall. Altman also highlighted a future request model upgrade, asking users whether a sole GPT‑4o or a new GPT‑4.5 would be preferable during the AMA.
Rollout and Stability Issues
OpenAI is currently permitting 3,000 advanced thinking queries per week for Pro users, while the platform remains slightly unstable during the rollout. The system has stabilized for Plus users (spending $200/month), but lower tier users still experience instability.
What’s Next?
OpenAI is poised to continue refining GPT‑5, encouraging users to experiment with different model variants and report any observable differences. This iterative approach underscores the brand’s commitment to professional refinement and user satisfaction.
