GPT-5 Bugs & Problems

Complete Issue Tracker - Updated December 2025

6,000+
Reddit Complaints at Launch
4,600
Upvotes on "GPT-5 is Horrible" Post
600+
Lines Before Code Breaks
"Bumpy"
Sam Altman's Own Words

Considering alternatives to ChatGPT? Compare top AI assistants to find options that prioritize reliability, privacy, and honest performance over hype.

Critical Bugs

Coding Capability Downgrade CRITICAL

GPT-5 breaks when generating complex code. Users report that scripts exceeding 600 lines consistently fail with variable initialization and scope issues.

"With GPT-5, it feels like the coding ability has been downgraded. Scripts that used to work now fail, solutions are weaker, and the model is less consistent."
— OpenAI Developer Community
Workaround:

Request GPT-4o instead (now restored for Plus users). Break code into smaller chunks under 500 lines.

Project Files Completely Broken CRITICAL

The Project Files feature in ChatGPT 5 is severely broken. Users report the AI cannot properly access or reference uploaded project files.

"Using Project Files in ChatGPT5 is very broke. It analyzes random code files instead of the ones I'm asking about, and generates suggestions for tasks I never requested."
— OpenAI Community Forum, December 2025
Autoswitcher Bug - "Way Dumber" Results CRITICAL

GPT-5's automatic model switching system breaks frequently, causing dramatically degraded responses.

"Yesterday, the autoswitcher broke and was out of commission for a chunk of the day, and the result was GPT-5 seemed way dumber."
— Sam Altman, OpenAI CEO

High Severity Issues

Persistent Hallucinations HIGH

Despite promises of reduced errors, GPT-5 still hallucinates at rates users find unacceptable for professional use. Real errors include wrong stock prices, confusing story details, and fabricated information.

"GPT-5 confidently told me incorrect stock prices, made up features that don't exist, and cited sources that were completely fabricated."
Basic Task Failures HIGH

Core capabilities that GPT-4o handled easily now fail. Basic summarization, organizing text into tables, and following simple instructions are unreliable.

"Things GPT-4o and o3 could handle, like basic summarization and organizing text into tables, GPT-5 stumbles through. It's a clear downgrade."
Output Throttling / Token Limiting HIGH

Many users believe GPT-5 is throttling token output to minimize costs, resulting in shorter, less helpful responses.

"Sounds like an OpenAI version of 'Shrinkflation'. Feels like cost-saving, not like improvement."
— Reddit User
Lost Personality / Creativity HIGH

GPT-5 is described as "creatively and emotionally flat" compared to GPT-4o. Users find it "genuinely unpleasant to talk to."

"Where GPT-4o could nudge me toward a more vibrant, emotionally resonant version of my own literary voice, GPT-5 sounds like a lobotomized drone."

Medium Severity Issues

File Download Failures MEDIUM

Generated files are frequently "no longer there" even when clicked immediately. The AI claims files will appear in 5 minutes but never delivers.

Chrome Hanging Issues MEDIUM

GPT-5 regularly hangs in Chrome browser, requiring page refreshes and losing conversation context.

Image Upload Recognition Failures MEDIUM

After uploading images, GPT-5 repeatedly asks users to re-upload, claiming it can't access files it just received.

OpenAI's Response

After massive backlash, OpenAI has:

However, many users report issues persist even after these fixes.

What Users Are Doing

Frustrated users are increasingly:

Related: Read more about performance decline trends →

Get the Full Report

Download our free PDF: "10 Real ChatGPT Failures That Cost Companies Money" (read it here) - with prevention strategies.

No spam. Unsubscribe anytime.

Need Help Fixing AI Mistakes?

We offer AI content audits, workflow failure analysis, and compliance reviews for organizations dealing with AI-generated content issues.

Request a consultation for a confidential assessment.