401–410 of 819 posts
Earlier quoted context omitted.
I get access to previews from OpenAI, Anthropic and Gemini pretty often. They're usually accompanied by an NDA and an embargo date - in this case the embargo was 10am Pacific this morning. I won't accept preview access if it comes with any conditions at all about what I can say about the model once the embargo has lifted.
[flagged]
its time to start benchmarking benchmarks. im pretty sure they are bmw levels doping the game here
I had access to a preview over the weekend, I published some notes here: https://simonwillison.net/2025/Sep/29/claude-sonnet-4-5/ It's very good - I think probably a tiny bit better than GPT-5-Codex, based on vibes more than a comprehensive comparison (there are plenty of benchmarks out there that attempt to be more methodical than vibes). It particularly shines when you try it on https://claude.ai/ using its brand n…
I am curious how the sandbox handles potentially malicious code. For example, what would happen if someone tried to run something like a crypto miner or a DDoS script?
If I was Anthropic I'd have measures in place to rate limit network traffic from individual containers.
Earlier quoted context omitted.
do you ever hit your pro quota?
Never hit pro quota yet, huge repo. Have multiple projects on the go locally and in cloud. Feel like this is going to be thr $1000 plan soon
When I see how much the latest models are capable of it makes me feel depressed. As well as potentially ruining my career in the next few years, its turning all the minutiae and specifics of writing clean code, that I've worked hard to learn over the past years, into irrelivent details. All the specifics I thought were so important are just implementation details of the prompt. Maybe I've got a fairly backwards view…
Earlier quoted context omitted.
That’s not the purpose of it, as I understand it; it’s a token phrase generated to cajole it down a particular path.[1] An alignment mechanism. The complement appears to be, “actually, that’s not right.”, a correction mechanism. 1: https://news.ycombinator.com/item?id=45137802
It gets annoying because A) it so quickly dismisses its own logic and conclusion from less than two minutes ago (extreme confidence with minimal conviction), and B) it fucks up the second time too (sometimes in the same way!) about 33% of the time.
Earlier quoted context omitted.
I'm not trying to be offensive here, feel the need to indicate that. But that prompt leads me to believe that you're going to get rather 'random' results due to leaving SO much room for interpretation. Also, in my experience, punctuation is important - particularly for pacing and grouping of logical 'parts' of a task and your prompt reads like a run on sentence. Making a lot of assumptions here - but I bet if I were…
But isn't the end goal to be able to get useful results without so much prompting? I mean in the movies for example, advanced AI assistants do amazing things with very little prompting. Seems like that's what people want. To me, the fact that so many people basically say "you are prompting it wrong" is knock against the tech and the model. If people want to say that these systems are so smart at what they can do, the…
Anecdotal evidence. I have a fairly large web application with ~200k LoC. Gave the same prompt to Sonnet 4.5 (Claude Code) and GPT-5-Codex (Codex CLI). "implement a fuzzy search for conversations and reports either when selecting "Go to Conversation" or "Go to Report" and typing the title or when the user types in the title in the main input field, and none of the standard elements match, a search starts with a 2s de…
And I actually like Claude more because of that.
codex will indeed work more independently but you will have hard time when it isn't what you want. It will use python script just to do simple edits in files ( niesearch and replace for unique code snippets in small files) when it's wrong good look convincing it (it already have some outdated info like on latest docker image releases and convincing it that Debian base changed was challenging) it uses context more effectively but it will lack explanation why it is doing what it is doing, asking it to explain will just cause it to do something else without any word.
and of course lack of proper permissions for running commands. sandbox is cool but I do not want it to be able to commit, I want it to just edit files or I want to have some more control over what it does.
you can run codex as mcp server, I prefer adding it to Claude and ask to do cooperative plan, codex will do great analysis and plan and I can comfortable work with Claude on the code that matches my style
There had been a trend where each new model released from OpenAI, Anthropic, etc. felt like a letdown or worse a downgrade.
But the release of 4.5 break that trend, And is a pleasant surprise on day one.
Well done! :)
When I see how much the latest models are capable of it makes me feel depressed. As well as potentially ruining my career in the next few years, its turning all the minutiae and specifics of writing clean code, that I've worked hard to learn over the past years, into irrelivent details. All the specifics I thought were so important are just implementation details of the prompt. Maybe I've got a fairly backwards view…