Wow GPT-Image-2 is so realistic
@petergostev
-
Faster generations enable potential Codex feature expansion
By
–
This is in ChatGPT, I'm just hypothesising that perhaps it would become feasible to add it to Codex now since the generations are faster – would be nice.
-
ChatGPT Pro Model: 3-4x Faster Generations with Enhanced Quality
By
–
The 'Pro' model in ChatGPT does look like a real upgrade – generations are 3-4x times faster and look at lot better. I wouldn't say it is a giant leap forward, but a material upgrade – generations are richer, with more detail and coherence.
— Peter Gostev (@petergostev) 19 avril 2026
Considering that generations take c.20… https://t.co/GwN98bUf8a pic.twitter.com/AYHtoQRDE4The 'Pro' model in ChatGPT does look like a real upgrade – generations are 3-4x times faster and look at lot better. I wouldn't say it is a giant leap forward, but a material upgrade – generations are richer, with more detail and coherence. Considering that generations take c.20
-
ChatGPT Pro Model Delivers Faster Generation and Excellent Quality
By
–
Pro model in ChatGPT does feel very different – the generations are a lot faster (20 mins vs 60-80 mins for Pro Extended) and the quality is really excellent.
— Peter Gostev (@petergostev) 19 avril 2026
I'll do a side by side later, but this golden gate is quite excellent vs what all other models can do in one shot. https://t.co/LnIYY9jUC7 pic.twitter.com/drXuztL7oZPro model in ChatGPT does feel very different – the generations are a lot faster (20 mins vs 60-80 mins for Pro Extended) and the quality is really excellent. I'll do a side by side later, but this golden gate is quite excellent vs what all other models can do in one shot.
-
TED Audience Skepticism Toward AI Technology
By
–
Wouldn't be surprised if TED audience are not that fond of AI
-
OpenAI leads reasoning capabilities, DeepSeek close behind
By
–
I feel like OpenAI is the only lab that really nailed reasoning. DeepSeek was probably the closest, but need to see a frontier model from them to be sure. Gemini's reasoning was quite weird and all over the place. Claude's reasoning never used to matter (non thinking were
-
Tokenmaxxer’s Secret Thoughts Before Sleep
By
–
Every tokenmaxxer's secret thoughts before they go to sleep at 4am
-
Data Viewer Tool Compares 130+ AI Model Variants
By
–
Data viewer is very nice you can see 130+ model variants yourself https://
petergpt.github.io/bullshit-bench
mark/viewer/index.v2.html
… -
Opus 4.7 Underperforms 4.6 on BullshitBench Testing
By
–
BullshitBench: Opus 4.7 did WORSE than Opus 4.6 family. The 'Max' thinking version did worse than non-thinking – 74% 'pushback' vs 83% for non-thinking. As always, code, data etc is on github
-
RAG Era: Evolution of Retrieval-Augmented Generation Technology
By
–
does feel like something from the RAG era