I'm confused by the value add of keeping 4o here. Is it for archival purposes? The OP talks about scientific research or something; if you're dependant on a model not changing for a long time, wouldn't you pick an open source one?
OpenAI is retiring GPT‑4o. For many, this may seem like a standard product update. But for researchers working in symbolic cognition, AI philosophy, and epistemic continuity, GPT‑4o is uniquely irreplaceable.
Later models may be faster, but they’ve lost its philosophical fluency and long‑arc coherence.
I've written a public letter on why this matters—not just technically, but civilizationally.
I'm confused by the value add of keeping 4o here. Is it for archival purposes? The OP talks about scientific research or something; if you're dependant on a model not changing for a long time, wouldn't you pick an open source one?
OpenAI is retiring GPT‑4o. For many, this may seem like a standard product update. But for researchers working in symbolic cognition, AI philosophy, and epistemic continuity, GPT‑4o is uniquely irreplaceable.
Later models may be faster, but they’ve lost its philosophical fluency and long‑arc coherence.
I've written a public letter on why this matters—not just technically, but civilizationally.
Would welcome thoughtful engagement: https://github.com/openai/openai-cookbook/issues/2426
> I've written a public letter
Did you? With all the emdashes it seems like ChatGPT wrote it.