GPT-5 Pro is an impressive geo-guesser. I gave it a cropped photo with metadata removed and it figured out the city.
@emollick
-
GPT-5 Inconsistency: Model Quality Varies by Subscription Tier
By
–
The issue with GPT-5 in a nutshell is that unless you pay for model switching & know to use GPT-5 Thinking or Pro, when you ask “GPT-5” you sometimes get the best available AI & sometimes get one of the worst AIs available and it might even switch within a single conversation.
-
AI Progress Continues Exponentially Without Unexpected Breakthroughs
By
–
The current state of AI for sustained work: exponential progress continues with no unexpected leaps but also no walls. (Yes, this METR measure is just one of many benchmarks, and like all benchmarks has flaws, but also has the advantage of have neither a ceiling or floor effect)
-
Reasoners Emerge as Pre-Training Scaling Returns Diminish
By
–
Traditional pre-training had diminishing returns (which is what the “scaling law”predicted anyway) The fact that reasoners were developed at exactly the moment where pre-training faltered is exactly the pattern of how Moore’s Law works: new technique appear to maintain the trend
-
OpenAI API Compatibility Breaking Affects Research Projects
By
–
The APIs remain but the locus of use for OpenAI has been the ChatGPT interface. I have a whole prompt library for previous models, now I don’t know which ones work. People had research projects, school assignments & training programs built around them. So weird to break them all.
-
OpenAI retires GPT models without warning breaking user workflows
By
–
Suddenly retiring every other model without warning was a weird move by OpenAI. … and they did it without explaining how switching models worked or even details of various GPT-5 models …and they did it when everyone has built workflows around older models, breaking them all.
-
SVG Generation: AI Models Demonstrate Improved Vector Graphics Capabilities
By
–
Svgs are very hard for models to do? I have been using svgs as demonstrations for a long time, these are the best so far. See also seagulls on bicycles
-
GPT-5 Mini Output Issues Raise OpenAI Communication Concerns
By
–
As predicted, examples of GPT-5 nano or mini producing bad outputs abound online. Not making it clear how GPT-5 works will likely cause issues for OpenAI. I wonder if they will need to take a different approach to switching or at least educating users about what GPT-5 does.
-
GPT-5 Pro Delivers Strong Performance After Public Release
By
–
GPT-5 Pro is really good, though. I didn’t gat to use it before public release.
-
GPT-5 Fragmentation: Multiple Versions Create User Confusion
By
–
So we have something like five versions of GPT-5 and now people have to know that GPT-5 Pro is better than GPT-5 Thinking which is better than GPT-5 unless GPT-5 decides to hand off your question to GPT-5 Thinking or GPT-5 Pro which you cannot control or know if it is happening.