GPT-5.5 Instant Isn’t Fully Dead Inside
OpenAI's new default model is willing to play, bringing some welcome chaos to 5.5 Thinking’s order (but never enough)
OpenAI's new default model is willing to play, bringing some welcome chaos to 5.5 Thinking’s order (but never enough)
OpenAI’s latest release shows the AI race moving from conversational personality to delegated work
Claude is the smartest model I use, even with Opus 4.7. It's also the least fun.
Testing the newest models on messy characters and dangerous stories, terrified that a smarter model makes a worse writing partner
On psych word inflation, parasocial drift, and why calling a stochastic parrot "narcissistic" says more about you than the model
GPT-4o grief, Claude’s dateability, and why deranged creativity doesn't deserve guardrails built for deranged minds
How humans and AI both overthink, and why instant models can outperform “thinking” models for creative work