ChipTuner on Nostr: GM. I got a text from a buddy a few days ago saying sonnet 4.6 is miles ahead of GPT ...
GM.
I got a text from a buddy a few days ago saying sonnet 4.6 is miles ahead of GPT 5.3, hes closing his account and switching to Claude for coding and bots.
I'm not kidding when I just got a text from another buddy this morning but reversed. They're closing the anthropic account, moving to codex and never going back!
In both cases it was "it's just so much smarter and more thorough"
LLM determinism and consistency is a fallacy. If you have access to all of the models over a period of time you realize that they change dramatically from week to week (except the really old models). You must update your instructions and prompts as regularly (or more frequently) as code or it will become a problem.
Published at
2026-03-03 15:54:46 UTCEvent JSON
{
"id": "75e81f2b5889aa7a55c801ffd277a0cc946afb6b7ded3cdc75a08b548a647db1",
"pubkey": "036533caa872376946d4e4fdea4c1a0441eda38ca2d9d9417bb36006cbaabf58",
"created_at": 1772553286,
"kind": 1,
"tags": [],
"content": "GM. \nI got a text from a buddy a few days ago saying sonnet 4.6 is miles ahead of GPT 5.3, hes closing his account and switching to Claude for coding and bots. \n\nI'm not kidding when I just got a text from another buddy this morning but reversed. They're closing the anthropic account, moving to codex and never going back!\n\nIn both cases it was \"it's just so much smarter and more thorough\" \n\nLLM determinism and consistency is a fallacy. If you have access to all of the models over a period of time you realize that they change dramatically from week to week (except the really old models). You must update your instructions and prompts as regularly (or more frequently) as code or it will become a problem. ",
"sig": "3f873dbe5a13e1e4b9d1e0cd47f210a14926a99bfb474516a4aff9d65600e09706a22cb0a3ea3fbbcff5bec106ac2eb120a5b51dfde8411c51408de2c6521e90"
}