Earlier quoted context omitted.
are you, by any chance, a bot?
Not at all. I was controlling the browser myself as a human, not via any bot.
Sorry this was a stupid joke
GPT-5.6 Sol Ultra will be in Codex
341–350 of 433 posts
Re: GPT-5.6 Sol Ultra will be in Codex
#342Earlier quoted context omitted.
I tried a pro model out the other day and thought there must have been a bug in Pi’s cost calculations. But no, it’s absolutely fucking insane. Wasn’t even any better at the task.
I really suspect that the models are basically the same below, it’s all in the prompt. The way I use them, surgically, they seem to perform about the same. Fable certainly hasn’t blow my socks off.
Yeah, the bigger models shine when it comes to complexity (making the right decisions regarding choices with second-order effects), ambiguity (esp. common sense) and time horizon (agentic steps and context size).
If your tasks are well defined and don't require a very large number of steps -- e.g. you're asking for small, clearly defined changes to the code -- you're fine with grok-4-fast. (Well, you would be fine if they hadn't killed it.)
I work in both of these modes, and I find that the latter actually benefits from dumber models, because smaller models are faster. The work shifts from async to realtime/interactive. So you can stay alert, keep track of what they're doing and iterate, instead of alt-tabbing, getting a coffee, and then spending extra time resynchronizing your mental model later.
Re: GPT-5.6 Sol Ultra will be in Codex
#343I 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.
Yes, I’ve noticed the same thing recently. Yesterday I burned through four 5-hour usage-limit resets in roughly four hours of wall-clock time. Same workflow, but it feels like I’m getting about half the mileage now, maybe even less.
Guess that means Sol is dropping soon..
Re: GPT-5.6 Sol Ultra will be in Codex
#344Earlier quoted context omitted.
Deepseek gave it away with R1. They did it again with V4 and DSpark. You don't use it because it's Chinese
I'm not American so I don't really care if it's Chinese or not. But at work I can only use the approved Enterprise Plans we have and we only have those with Anthropic and OpenAI.
The shift may come when your blessed cloud platform provides competitive enterprise-ready hosting of the oss models
Re: GPT-5.6 Sol Ultra will be in Codex
#345Earlier quoted context omitted.
I reject the premise that using LLMs absolutely leads to loss of institutional knowledge. It is trivial for an LLM to generate a knowledge base of any kind in any language which can answer any question about your institution at any time. How is a bunch of fragmented humans with limited knowledge who can’t all communicate with each other better than that?
Have you ever taken the time to read hundreds of pages of documentation to fully understand a massive codebase? Neither have I. You learn it by working with it all day, and you're careful with it. A complete "knowledge base" of business logic is, itself, indistinguishable from code. Code that no one can or will read and learn unless they have to be immersed in it. So the "knowledge base" an LLM generates is not usefu…
> create a reverse payment instead of voiding a charge
I think I know the reason to this one. Maybe. Because the money is already gone! There are no rollbacks in the banking system only "counter"-transactions.
Re: GPT-5.6 Sol Ultra will be in Codex
#346Earlier quoted context omitted.
Still avialable through the API. According to people that have tried both Fable nad 5.6, Fable is clearly better at coding. So i expect a lot of people to pay extra for it.
But if 5.6 is better than Opus, Claude Code Max plan users will switch to OpenAI Codex en masse. Using Fable at API pricing is expensive
You’re not kidding. I ran out of Fable tokens a couple days ago, loaded up $70 just to finish the task we were working on, and burned through it in about 15 minutes.
Aaaaand it’s gone.
Re: GPT-5.6 Sol Ultra will be in Codex
#347Earlier quoted context omitted.
> If the Model makes repeated mistakes on the same subject matter, you can update your agent.md file ... That's all just prompting. > How do you think models are created? They are trained on feedback and learn. No one is post training models on a single mistake. At least I have not seen it. I also doubt it is effective. Post-training on a single failure will not meaningfully change the model. That even sidesteps the…
> That's all just prompting And telling someone not to repeat a mistake is… ?
Are you really making the case that teaching a person how to work is equivalent to prompting an AI?
Re: GPT-5.6 Sol Ultra will be in Codex
#348there seems to be very big misunderstanding about what the "ultra" is, so let me explain it basing on the codex source code: it's similar to Claude code ultracode. there is no ultra effort level implemented on the backend. it's just alias in the codex to max effort setting and single line addition to prompt to use subagents proactively. that's all as far as we know pro models work differently. for once those are back…
> single line addition to prompt to use subagents proactively. This misses an important detail. In Claude Code [1], ultracode suggests the agent create a JavaScript code to deterministically orchestrate sub agents. This is different from just having the main agent launch sub agents and (non-deterministically) manage them. The resulting workflow is called “dynamic” because CC creates this orchestration script dynamica…
Can you explain what you find useful about ultracode? I've become wary of agent swarms since the early days and now just prefer to have a single agent spin for hours at time. Parallelism never got me anywhere worthwhile.
Re: GPT-5.6 Sol Ultra will be in Codex
#349Earlier quoted context omitted.
[flagged]
Remove the #m at end of the link. That's causing the redirect. Here is a cleaner link: https://xcancel.com/haider1/status/2073695124220006575
I tried that, and no, it didn't help. The problem persists. Even a private window doesn't help. I suspect Firefox is just shitlisted by them.
Re: GPT-5.6 Sol Ultra will be in Codex
#350Earlier quoted context omitted.
Ok I’m not sure I follow your point here. Isn’t all that he’s saying that if they find some optimization techniques, that gives them an edge? And that makes sense? How is this suddenly evidence of him being a villain?
The evidence is that unlike Deepseek he does not publish his compute multipliers. Under that argument Deepseek should not publish any of their research either.
DeepSeek is acting in accordance with their incentives, just like Anthropic is