Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Claude Opus 4.7

anthropic.com

Post reply on HN
Print view View original

191–200 of 1001 posts

Re: Claude Opus 4.7

#191

Post by fluidcruft »

alvis wrote: post #19
buildbot wrote: post #12

Too late, personally after how bad 4.6 was the past week I was pushed to codex, which seems to mostly work at the same level from day to day. Just last night I was trying to get 4.6 to lookup how to do some simple tensor parallel work, and the agent used 0 web fetches and just hallucinated 17K very wrong tokens. Then the main agent decided to pretend to implement tp, and just copied the entire model to each node...

I don't have much quality drop from 4.6. But I also notice that I use codex more often these days than claude code

I generally think codex is doing well until I come in with my Opus sweep to clean it up. Claude just codes closer to the way my brain works. codex is great at finding numerical stability issues though and increasingly I like that it waits for an explicit push to start working. But talking to Claude Code the way I learned to talk to codex seems to work also so I think a lot of it is just learning curve (for me).

Re: Claude Opus 4.7

#193

Post by velcrovan »

Kim_Bruning wrote: post #3

> "We are releasing Opus 4.7 with safeguards that automatically detect and block requests that indicate prohibited or high-risk cybersecurity uses. " This decision is potentially fatal. You need symmetric capability to research and prevent attacks in the first place. The opposite approach is 'merely' fraught. They're in a bit of a bind here.

Questions about "fatality" aside, where do you see asymmetry here?

Re: Claude Opus 4.7

#194

Post by corlinp »

I'm running it for the first time and this is what the thinking looks like. Opus seems highly concerned about whether or not I'm asking it to develop malware.

> This is _, not malware. Continuing the brainstorming process.

> Not malware — standard _ code. Continuing exploration.

> Not malware. Let me check front-end components for _.

> Not malware. Checking validation code and _.

> Not malware.

> Not malware.

Re: Claude Opus 4.7

#195

Post by onlyrealcuzzo »

buildbot wrote: post #12

Too late, personally after how bad 4.6 was the past week I was pushed to codex, which seems to mostly work at the same level from day to day. Just last night I was trying to get 4.6 to lookup how to do some simple tensor parallel work, and the agent used 0 web fetches and just hallucinated 17K very wrong tokens. Then the main agent decided to pretend to implement tp, and just copied the entire model to each node...

I switched to Codex and found it extremely inferior for my use case.

It is much faster, but faster worse code is a step in the wrong direction. You're just rapidly accumulating bugs and tech debt, rather than more slowly moving in the correct direction.

I'm a big fan of Gemini in general, but at least in my experience Gemini Cli is VERY FAR behind either Codex or CC. It's both slower than CC, MUCH slower than Codex, and the output quality considerably worse than CC (probably worse than Codex and orders of magnitude slower).

In my experience, Codex is extraordinarily sycophantic in coding, which is a trait that could t be more harmful. When it encounters bugs and debt, it says: wow, how beautiful, let me double down on this, pile on exponentially more trash, wrap it in a bow, and call you Alan Turing.

It also does not follow directions. When you tell it how to do something, it will say, nah, I have a better faster way, I'll just ignore the user and do my thing instead. CC will stop and ask for feedback much more often.

YMMV.

Re: Claude Opus 4.7

#196

Post by _the_inflator »

buildbot wrote: post #12

Too late, personally after how bad 4.6 was the past week I was pushed to codex, which seems to mostly work at the same level from day to day. Just last night I was trying to get 4.6 to lookup how to do some simple tensor parallel work, and the agent used 0 web fetches and just hallucinated 17K very wrong tokens. Then the main agent decided to pretend to implement tp, and just copied the entire model to each node...

Codex really has its place in my bag. I mainly use it, rarely Claude.

Codex just gets it done. Very self-correcting by design while Claude has no real base line quality for me. Claude was awesome in December, but Codex is like a corporate company to me. Maybe it looks uncool, but can execute very well.

Also Web Design looks really smooth with Codex.

OpenAI really impressed me and continues to impress me with Codex. OpenAI made no fuzz about it, instead let results speak. It is as if Codex has no marketing department, just its product quality - kind of like Google in its early days with every product.

Re: Claude Opus 4.7

#197

Post by orrito »

mbeavitt wrote: post #34

Honestly I've been doing a lot of image-related work recently and the biggest thing here for me is the 3x higher resolution images which can be submitted. This is huge for anyone working with graphs, scientific photographs, etc. The accuracy on a simple automated photograph processing pipeline I recently implemented with Opus 4.6 was about 40% which I was surprised at (simple OCR and recognition of basic features). I…

Did you try the same with gemini 3 models? Those usually score higher on vision benchmarks

Re: Claude Opus 4.7

#198

Post by mwigdahl »

jimmypk wrote: post #81

[flagged]

That depends a bit on token efficiency. From their "Agentic coding performance by effort level" graph, it looks like they get similar outcome for 4.7 medium at half the token usage as 4.6 at high.

Granted that is, as you say, a single prompt, but it is using the agentic process where the model self prompts until completion. It's conceivable the model uses fewer tokens for the same result with appropriate effort settings.

Re: Claude Opus 4.7

#200

Post by lanyard-textile »

This comment thread is a good learner for founders; look at how much anguish can be put to bed with just a little honest communication.

1. Oops, we're oversubscribed.

2. Oops, adaptive reasoning landed poorly / we have to do it for capacity reasons.

3. Here's how subscriptions work. Am I really writing this bullet point?

As someone with a production application pinned on Opus 4.5, it is extremely difficult to tell apart what is code harness drama and what is a problem with the underlying model. It's all just meshed together now without any further details on what's affected.

Post reply on HN