With encouragement to actually think, it gets sooo close!
@jeremyphoward
-
Nine-Year-Old Outperforms GPT5 on Reasoning Task
By
–
My 9yo got it straight away. GPT5 on ChatGPT gets it totally wrong.
-
Codex CLI Reliability Issues Compared to Claude Code
By
–
Ugh codex CLI seems a lot flakier than claude code… 🙁
-
ChatGPT Canvas GPT5 Multi-Step Tasks Broken Issue
By
–
Actually, upon further testing, it turns out that trying to do almost anything multi-step in ChatGPT's canvas with GPT5 is totally broken. Will try codex and see if that works.
-
GPT5 thinking compares well with Claude Opus router issues
By
–
GPT5 thinking seems pretty good – up there with the latest Claude Opus AFAICT. But the router model on ChatGPT seems to have been trained to route far too much to the nerfed versions.
-
Add Think Hard to GPT5 Prompts for Better Performance
By
–
PSA: Add ". think hard" to the end of all your ChatGPT GPT5 prompts. In my testing, so far that has resulted in it using the competent model 100% of the time. And so far, not adding it has resulted in it using the crippled model 100% of the time, which has failed all my tasks.
-
Perl: Powerful Alternative to Bash Scripting Tools
By
–
No lol! :O Perl is legit awesome for anything beyond the most basic bash script. And it's available everywhere. You can get rid of awk, find, and sed, and make life so much easier 😀
-
How Does Sonnet 4 Perform in AI Model Comparisons?
By
–
Thanks for sharing these really helpful results. 🙂 how does Sonnet 4 do?
-
20-Hour Problem Solving Course Available at fast.ai
By
–
That's a very long story – we ran a >20 hour course on it (as part of a general approach to problem solving) http://
solveit.fast.ai There will likely be another in the next couple of months if you're interested! 😀