Why everyone HATES GPT-5 (and how to fix it)
Related videos
The Truth About My Channel (And How Much $$ I Make)
Matt Wolfe
20.2k views
This SHOCKING New Bill Is Why Everyone Hates Congress | Ep. 1764
Matt Walsh
73.0k views
AI News: Everyone's Leaving ChatGPT!
Matt Wolfe
54.4k views
The AI Stack That Saves Hours Every Day
Matt Wolfe
37.4k views
The AI Wake-Up Call Everyone Needs Right Now!
Matt Wolfe
88.5k views
How To Build AI Agents To Automate EVERYTHING (n8n tutorial)
Matt Wolfe
74.3k views
Complete ChatGPT-5 Breakdown and First Impressions
Matt Wolfe
203.8k views
The history of AI that no one ever told you
Matt Wolfe
43.3k views
Why everyone is freaking out about Veo 3, Claude 4 & OpenAI’s iPhone moment
Matt Wolfe
145.2k views
Every Major AI Update This Week in One Video
Matt Wolfe
103.1k views
Evaluating User Gripes Against the Real-World Performance of ChatGPT-5
Discover which major criticisms leveled against the ChatGPT-5 launch—from missing models to poor coding—are still valid today versus those that OpenAI already fixed. Readers gain immediate clarity on GPT-5’s current strengths and weaknesses compared to legacy models.
Short Summary
- Restore access to legacy models like GPT-4o by manually enabling the "Show legacy models" setting.
- Initial performance dips for personality and general accuracy were significantly influenced by a broken automatic model router on launch day.
- Specific coding tests indicate GPT-5 currently lags behind competitors like Claude Opus 4.1 and even GPT-3.5 Pro in producing functional code snippets.
- Accuracy concerning logic puzzles and math remains inconsistent, requiring users to verify responses closely rather than blindly trusting the output.
This analysis deep-dives into four major complaints following the GPT-5 release: model availability, personality, coding ability, and accuracy. We test these facets to validate user frustration and identify recent patches or workarounds implemented by OpenAI, helping you integrate the new model effectively into your workflow.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
The Truth About My Channel (And How Much $$ I Make)
Matt Wolfe
20.2k views
This SHOCKING New Bill Is Why Everyone Hates Congress | Ep. 1764
Matt Walsh
73.0k views
AI News: Everyone's Leaving ChatGPT!
Matt Wolfe
54.4k views
The AI Stack That Saves Hours Every Day
Matt Wolfe
37.4k views
The AI Wake-Up Call Everyone Needs Right Now!
Matt Wolfe
88.5k views
How To Build AI Agents To Automate EVERYTHING (n8n tutorial)
Matt Wolfe
74.3k views
Complete ChatGPT-5 Breakdown and First Impressions
Matt Wolfe
203.8k views
The history of AI that no one ever told you
Matt Wolfe
43.3k views
Why everyone is freaking out about Veo 3, Claude 4 & OpenAI’s iPhone moment
Matt Wolfe
145.2k views
Every Major AI Update This Week in One Video
Matt Wolfe
103.1k views
Top Comments (10)
GPT-5 has been 10x worse for me when it comes to writing or structuring research. It's like it's actively trying to piss you off and ignores all specific instructions.
GPT-5 is the worst model I’ve ever tried. It’s driving me insane.. I feel like smashing my computer. I want the old models back!
Just spent the last hour trying to get it to do one thing. It kept saying "Okay, I'll do that now, then it would ask me another question and I say yes, and it would say okay I will do it now, and then ask me another dumb question, and then say on the next one I will do it, and then say right after this I will do, but NEVER DO IT!!! it was a simple task, and it had done it ones I just wanted to update it. I truly believe there is someone watching to see what people's boiling points are, getting a laugh out of pissing people off with this. I pay for a service, but I'm not getting the service.
For me, 4o is much better at writing than GPT-5.
I never used curse words in chat gpt until gpt 5
I have to chime in on the "personality gripe" stuff because hello, that was a FANTASTIC answer that 4o gave for the "cheer me up" prompt. Are you kidding? That was a night-and-day comparison that I think encapsulates the exact personality shift that people are griping about. GPT5's response was like your best friend giving you a card when you're down without writing anything in it. You open it up and there's just some crap about otters that FOOD4LESS wrote. Instant suicide right there. 4o actually took the prompt seriously. You can tell because it used the entire spectrum of its intelligence; Emotional reasoning, context, COMMON SENSE! But then it decided to not make it about itself and what it can do, but rather to cut the shit and speak to Matt personally. That was LITERALLY the best response possible. There's none above it. Gripe valid all the way.
I have never typed ok so many times ,it asks me more questions than I ask it
I end up adding “DO WHAT THE FUCK I ASK AND DONT ADD ANY OTHER BULLSHIT” to my prompts
The only reason choosing a model is confusing is because we have to choose from arbitrary names. Imagine if the model names were actually descriptive as to what the model was best suited for? "Conversational Model", "Coding Model", "Diagnostic Model", etc
Gpt 5 is like talking to angry condescending HR staff
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
GPT-5 has been 10x worse for me when it comes to writing or structuring research. It's like it's actively trying to piss you off and ignores all specific instructions.
GPT-5 is the worst model I’ve ever tried. It’s driving me insane.. I feel like smashing my computer. I want the old models back!
Just spent the last hour trying to get it to do one thing. It kept saying "Okay, I'll do that now, then it would ask me another question and I say yes, and it would say okay I will do it now, and then ask me another dumb question, and then say on the next one I will do it, and then say right after this I will do, but NEVER DO IT!!! it was a simple task, and it had done it ones I just wanted to update it. I truly believe there is someone watching to see what people's boiling points are, getting a laugh out of pissing people off with this. I pay for a service, but I'm not getting the service.
For me, 4o is much better at writing than GPT-5.
I never used curse words in chat gpt until gpt 5
I have to chime in on the "personality gripe" stuff because hello, that was a FANTASTIC answer that 4o gave for the "cheer me up" prompt. Are you kidding? That was a night-and-day comparison that I think encapsulates the exact personality shift that people are griping about. GPT5's response was like your best friend giving you a card when you're down without writing anything in it. You open it up and there's just some crap about otters that FOOD4LESS wrote. Instant suicide right there. 4o actually took the prompt seriously. You can tell because it used the entire spectrum of its intelligence; Emotional reasoning, context, COMMON SENSE! But then it decided to not make it about itself and what it can do, but rather to cut the shit and speak to Matt personally. That was LITERALLY the best response possible. There's none above it. Gripe valid all the way.
I have never typed ok so many times ,it asks me more questions than I ask it
I end up adding “DO WHAT THE FUCK I ASK AND DONT ADD ANY OTHER BULLSHIT” to my prompts
The only reason choosing a model is confusing is because we have to choose from arbitrary names. Imagine if the model names were actually descriptive as to what the model was best suited for? "Conversational Model", "Coding Model", "Diagnostic Model", etc
Gpt 5 is like talking to angry condescending HR staff