The full conversation https://xcancel.com/haider1/status/2073695124220006575#m
[flagged]
are you, by any chance, a bot?
GPT-5.6 Sol Ultra will be in Codex
31–40 of 433 posts
Re: GPT-5.6 Sol Ultra will be in Codex
#32Earlier quoted context omitted.
Like google search, this does not work because of how common long tail use is. What you think could be a big chunk, is more likely to be a fraction of a percent of queries. And what use is similar query caching - so you (very often! if actually cost effective, maybe half the time) get a response to a query that was different from yours. Including for when you have a lot of context input already. You’re going to get t…
I would be very surprised if they hadn’t sorted out some form of shared KV caching
I wouldn't
Re: GPT-5.6 Sol Ultra will be in Codex
#33Earlier quoted context omitted.
Like google search, this does not work because of how common long tail use is. What you think could be a big chunk, is more likely to be a fraction of a percent of queries. And what use is similar query caching - so you (very often! if actually cost effective, maybe half the time) get a response to a query that was different from yours. Including for when you have a lot of context input already. You’re going to get t…
Please pardon the pure speculation incoming. Yes, caching the answer doesn't seem useful. Caching the progression, the graph, may be. This is similar to making code changes with ed(1) instead of editing in vi. The transform script(s) are cached and can be played back or adjusted. Surely for some breadth of question inputs, they map more often to similar answers--but not static answers; instead, evented edits. It's ne…
I don't understand how this fits LLM architecture at all
Re: GPT-5.6 Sol Ultra will be in Codex
#34Earlier quoted context omitted.
Maybe cache similar answers from others. Surprised if this is not already being done.
Like google search, this does not work because of how common long tail use is. What you think could be a big chunk, is more likely to be a fraction of a percent of queries. And what use is similar query caching - so you (very often! if actually cost effective, maybe half the time) get a response to a query that was different from yours. Including for when you have a lot of context input already. You’re going to get t…
But there must be a ton of generic questions that people ask. Stuff like "What's the capital of country X?" - it's probably at least 10% of queries. Memories, custom instructions etc would invalidate them, but if you can return the answers basically free it's probably worth it.
Re: GPT-5.6 Sol Ultra will be in Codex
#35I wonder if it's related that that OpenAI has found a way to cut inference costs by half, according to The Information. https://www.theinformation.com/newsletters/ai-agenda/openai-...
Semi-related, has anyone noticed their GPT 5.5 usage in Codex being cut in half as of a couple days ago? I got a lot more mileage out of my session usage yesterday for the same workload.
Re: GPT-5.6 Sol Ultra will be in Codex
#36Earlier quoted context omitted.
Is it as good as Fable..? Fable is the first model that mostly writes without the AI slop format for me, and so I can comfortably actually copy and paste most of what it spits out. OpenAI models have always been the worst in my experience for verbose, slop formatted responses, with each generation increasing in sloppiness.
> Fable is the first model that mostly writes without the AI slop format for me I'm not that impressed by Fable's writing to be honest, still has the AI giveaways like em dash.
Humans use em dash as well.
I hate that I have had to remove it from my writing style because people assume it’s AI generated. But I think that ship has sailed. I’ll have to do without now.
Re: GPT-5.6 Sol Ultra will be in Codex
#37Recently, I've been so eager to get new model releases in Codex. I'm hooked. I hope this accelerates development. Shows how dependant I have become to Codex.
Re: GPT-5.6 Sol Ultra will be in Codex
#38Earlier quoted context omitted.
Is it as good as Fable..? Fable is the first model that mostly writes without the AI slop format for me, and so I can comfortably actually copy and paste most of what it spits out. OpenAI models have always been the worst in my experience for verbose, slop formatted responses, with each generation increasing in sloppiness.
> Fable is the first model that mostly writes without the AI slop format for me I'm not that impressed by Fable's writing to be honest, still has the AI giveaways like em dash.
i cant reply to hn_user2, but i have the same experience, i find myself never using emdash where i would have before
Re: GPT-5.6 Sol Ultra will be in Codex
#39No Twitter, what’s he responding to?
Check out LibRedirect or Predirect (MV3), it automatically redirects youtube, X, etc links to privacy-respecting frontends.
https://news.ycombinator.com/item?id=44344246
107 comments, 1 year ago.
Re: GPT-5.6 Sol Ultra will be in Codex
#40Earlier quoted context omitted.
For pro mode the agents worked independently and only when they all finished did a new agent take a look at everything to merge the work into a single response. The new thing involves subagents that have been trained to cooperatively pursue a task and are allowed to communicate with each other along the way.
Do you have a source for this, or just rumors? The responses I get from pro don't feel like ensembles. They are often very one directional.
This can be because the summary model just picked the output from one of the sub agents.