GPT-4o

4o had some notable problems with sycophancy being very very positive about the user and going along with almost anything the user said. OpenAI even talked about it [0] and the new responses to people trying to continue their former 'relationship' does tend towards being 'harsh' [1] especially if you were a person actually thinking of the bot as a kind of person. [0] https://openai.com/index/sycophancy-in-gpt-4o/ [1] https://www.reddit.com/r/MyBoyfriendIsAI/comments/1qx3jux/wh

I noticed that LLMs like to write code and anytime an "AI feature" is needed they will heavily default to using `gpt-4o` as kind of the "hello world" of models. It was a good model when it came out and a lot of people started building on it, which caused the training data to be saturated by it. My AGENTS.md has: You MUST use a modern but cost effective LLM such as `qwen3-8b` when you need structured output or tool support. The reality is that almost all LLMs have quirks and each provider tries their best to smooth them over, but often you might start seeing stuff specific to OpenAI or the `gpt-4o` model in the code. IMO the last thing you want to be doing in 2026 is paying higher costs to use an outdated model being kept on life support that needs special tweaks that won't be relevant once it gets the axe.

> ChatGPT’s new image generation in GPT‑4o rolls out starting today to Plus, Pro, Team, and Free users as the default image generator in ChatGPT, with access coming soon to Enterprise and Edu. For those who hold a special place in their hearts for DALL·E, it can still be accessed through a dedicated DALL·E GPT. > Developers will soon be able to generate images with GPT‑4o via the API, with access rolling out in the next few weeks. That's it folks. Tens of thousands of so-called "AI" image generator startups have been obliterated and taking digital artists with them all reduced to near zero. Now you have a widely accessible meme generator with the name "ChatGPT". The last task is for an open weight model that competes against this and is faster and all for free.

I hesitated to share this because I’m not a typical HN user, but here’s my experience with GPT‑4o… I’m not a technical user. I studied humanities, and I mostly work in public service. GPT‑4o has been quietly transformative for me in day-to-day life, not because it’s flashy, but because it helps with things I couldn’t ask a person to do over and over again without guilt or shame. When I experience depressive episodes or social anxiety, I lose the ability to process thoughts clearly or communicate well. GPT‑4o helps me organize my thoughts, write messages with clarity and calm, and prepare myself mentally for difficult conversations. It doesn’t “replace” humans — but it helps me stay connected with the world when I’m at my lowest. I’ve also used it as a mirror to think more deeply about myself — not in a therapeutic sense, but in a reflective one. I’ve journaled with it, questioned my values with it, and tried to understand parts of myself that otherwise remain buried in noise or self-censorship. It’s strange to say, but the consistent tone and memory helped me build internal continuity, especially during periods when I didn’t feel like “myself.” To me, this tool has been part cognitive scaffold, part co‑writer, part emotional stabilizer. That’s not just sentimentalism — it affects how well I function. These subtle, personal use cases are rarely seen in blog posts or launch demos, but I hope they are part of the conversation. Especially for those of us outside the US tech bubble, we often don’t get to “vote” except through usage. And when something this impactful quietly disappears, we feel powerless. I’m not asking to halt progress. Just that, if possible, people making decisions about AI products also hear stories like this — of how these tools don’t just perform, but support.