Early impression of #OpenAI's O1 model: I'm very much still chatting with an LLM, with all that this implies, but it's better at self-guiding towards the kind of quality output that would ordinarily take retries/coaxing, at the cost of taking longer to respond (and of course using more energy). Compared with #Claude, it actually did worse at a couple of programming challenges I threw at it, but I suspect on average it'll benchmark a step above it.
#AI
Erik Moeller /
npub166…m7wr3
2024-09-12 22:12:12
Author Public Key
npub16697q673apk05pjejpge96d8td2eu09efluusv44fkt6a8efs46srm7wr3Published at
2024-09-12 22:12:12Event JSON
{
"id": "cbe13f639f92d58cbbd627e881387fac95a01430ff993c72cdd00501da2e54cd",
"pubkey": "d68be06bd1e86cfa0659905192e9a75b559e3cb94ff9c832b54d97ae9f298575",
"created_at": 1726179132,
"kind": 1,
"tags": [
[
"t",
"openai"
],
[
"t",
"claude"
],
[
"t",
"ai"
],
[
"proxy",
"https://social.coop/users/eloquence/statuses/113126875634956104",
"activitypub"
]
],
"content": "Early impression of #OpenAI's O1 model: I'm very much still chatting with an LLM, with all that this implies, but it's better at self-guiding towards the kind of quality output that would ordinarily take retries/coaxing, at the cost of taking longer to respond (and of course using more energy). Compared with #Claude, it actually did worse at a couple of programming challenges I threw at it, but I suspect on average it'll benchmark a step above it.\n\n#AI",
"sig": "8862b702eef2ccef6b77c165bc5fca4a39029be72a90027609551f4c359ff049a2e33145ad5ca57096554675f55e3004dc90fbc2060f75c60b887a9e6e4f13d5"
}