Uh Oh! GPT-5 Getting 'Horrible' Reviews

OpenAI’s rollout of GPT-5 — now the default model for ChatGPT — has ignited a wave of subscriber backlash. Many paying users say the new model is slower, less accurate, and more restrictive than the GPT-4o and 4.1 options it replaces.

Key Complaints

• Performance concerns: Reports of GPT-5 producing weaker or less precise answers than older models.
• Loss of model choice: Removal of model selection forces a one-size-fits-all approach.
• Usage restrictions: GPT-5 Thinking mode is capped at 200 messages per week for ChatGPT Plus subscribers.

Community Outcry

A Reddit thread titled "GPT-5 is horrible" drew over 3,000 upvotes within 24 hours, with more than 1,200 comments. Top posts include:

“I miss 4.1. Bring it back.”
“Their demo basically said ‘just keep retrying until you get a good answer’ — how is this progress?”
“This feels like AI shrinkflation — we’re paying more for less.”

The Expectations Gap

The criticism follows a high-profile pre-launch campaign from OpenAI CEO Sam Altman, who:

• Teased the update with Death Star imagery on X (formerly Twitter).
• Promised PhD-level reasoning capabilities.
• Positioned GPT-5 as a transformative leap forward.

Instead, many users say the upgrade feels incremental — and in some cases, like a downgrade.

OpenAI’s Position

OpenAI maintains that GPT-5:

• Dynamically adapts its reasoning abilities without requiring multiple specialized models.
• Outperforms earlier versions on benchmark tests.
• Represents the future of conversational AI.

Ongoing Issues

Despite the company’s claims, users continue to report:

• Slower response times.
• Inconsistent answer quality.
• Frustration over losing access to more reliable, older models.

Industry analysts warn that unless OpenAI addresses these issues quickly, rivals such as Anthropic and Google’s Gemini could capitalize on the discontent.