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

671–680 of 1001 posts

Re: Claude Opus 4.7

#671

Post by SkyPuncher »

lanyard-textile wrote: post #200

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 ap…

I agree.

I have flexibility to shift my core working hours (and what I do during N/A business hours). Knowing they're explicitly making it dumb because of load is important. It allows me to shuffle my work around and run heavy workloads late at night (plan during working hours then come click "yes" a few times in the evening).

Re: Claude Opus 4.7

#672

Post by davesque »

earthnail wrote: post #664
davesque wrote: post #662

> We stated that we would keep Claude Mythos Preview’s release limited and test new cyber safeguards on less capable models first. Opus 4.7 is the first such model: its cyber capabilities are not as advanced as those of Mythos Preview (indeed, during its training we experimented with efforts to differentially reduce these capabilities). We are releasing Opus 4.7 with safeguards that automatically detect and block req…

I feel it’s fine as a short term solution, and probably a good thing. Gives the good guys some time to stay on top. Always remember: a defender must succeed every time , an attacker only once.

So then why expect that you're making the world safer by limiting the capability that your vendor locked customers have access to while attackers will go find the best de-censored model that works for them, wherever they can find it?

Re: Claude Opus 4.7

#673

Post by andai »

MarkMarine wrote: post #516
puppystench wrote: post #364

Earlier quoted context omitted.

Does this mean Claude no longer outputs the full raw reasoning, only summaries? At one point, exposing the LLM's full CoT was considered a core safety tenet.

Anthropic was chirping about Chinese model companies distilling Claude with the thinking traces, and then the thinking traces started to disappear. Looks like the output product and our understanding has been negatively affected but that pales in comparison with protecting the IP of the model I guess.

When Gemini Pro came out, I found the thinking traces to be extremely valuable. Ironically, I found them much more readable than the final output. They were a structured, logical breakdown of the problem. The final output was a big blob of prose. They removed the traces a few weeks later.

Re: Claude Opus 4.7

#674

Post by andai »

davesque wrote: post #662

> We stated that we would keep Claude Mythos Preview’s release limited and test new cyber safeguards on less capable models first. Opus 4.7 is the first such model: its cyber capabilities are not as advanced as those of Mythos Preview (indeed, during its training we experimented with efforts to differentially reduce these capabilities). We are releasing Opus 4.7 with safeguards that automatically detect and block req…

The fundamental tension is that the models are getting weirdly good at hacking while still sort of sucking at a bunch of economically valuable tasks.

So they've hit the point where the models are simultaneously too smart (dangerous hacking abilities) and too stupid (can't actually replace most employees). So at this point they need to make the models bigger, but they're already too big.

So the only thing left to do is to make them selectively stupider. I didn't think that would be possible, but it seems like they're already working on that.

Re: Claude Opus 4.7

#675

Post by gertlabs »

OsrsNeedsf2P wrote: post #667
gertlabs wrote: post #629

Early benchmark results on our private complex reasoning suite: https://gertlabs.com/?mode=agentic_coding Opus 4.7 is more strategic, more intelligent, and has a higher intelligence floor than 4.6 or 4.5. It's roughly tied with GPT 5.4 as the frontier model for one-shot coding reasoning, and in agentic sessions with tools, it IS the best, as advertised (slightly edging out Opus 4.5, not a typo). We're still running m…

Do your benchmark results indicate any level of regression on Opus 4.6 or 4.5 since their first release?

We only have some basic time filtering (https://gertlabs.com/?days=30), but most of our samples are from the last 2 months. This is a visualization we plan to add when we've collected more historical data.

But we did heavily resample Claude Opus 4.6 during the height of the degraded performance fiasco, and my takeaway is that API-based eval performance was... about the same. Claude Opus 4.6 was just never significantly better than 4.5.

But we don't really know if you're getting a different model when authenticated by OAUTH/subscription vs calling the API and paying usage prices. I definitely noticed performance issues recently, too, so I suspect it had more to do with subscription-only degradation and/or hastily shipped harness changes.

Re: Claude Opus 4.7

#676

Post by russellthehippo »

Initial testing today - 4.7 excels at abstractions/implementations of abstractions in ways that often failed in 4.5/4.6. This is a great update, I've had to do a lot of manual spec to ensure consistency between design and implementation recently as projects grow.

Re: Claude Opus 4.7

#677

Post by andai »

ayewo wrote: post #463
johnmlussier wrote: post #227

They've increased their cybersecurity usage filters to the point that Opus 4.7 refuses to work on any valid work, even after web fetching the program guidelines itself and acknowledging "This is authorized research under the [Redacted] Bounty program, so the findings here are defensive research outputs, not malware. I'll analyze and draft, not weaponize anything beyond what's needed to prove the bug to [Redacted]. I…

Sounds like you will need to drink a(n identity) verification can soon [1] to continue as a security researcher on their platform. 1: https://support.claude.com/en/articles/14328960-identity-ver... Identity verification on Claude Being responsible with powerful technology starts with knowing who is using it. Identity verification helps us prevent abuse, enforce our usage policies, and comply with legal obligations. W…

Context for "please drink verification can": https://files.catbox.moe/eqg0b2.png

Re: Claude Opus 4.7

#678

Post by andai »

zarzavat wrote: post #336
lanyard-textile wrote: post #200

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 ap…

These threads are always full of superstitious nonsense. Had a bad week at the AIs? Someone at Anthropic must have nerfed the model! The roulette wheel isn't rigged, sometimes you're just unlucky. Try another spin, maybe you'll do better. Or just write your own code.

They don't nerf the model, just lower the default reasoning effort, encourage shorter responses in the system prompt, etc. Totally different ;)

Re: Claude Opus 4.7

#679

Post by andai »

trueno wrote: post #455

noticing sharp uptick in "i switched to codex" replies lately. a "codex for everything" post flocking the front page on the day of the opus 4.7 release me and coworker just gave codex a 3 day pilot and it was not even close to the accuracy and ability to complete & problem solve through what we've been using claude for. are we being spammed? great. annoying. i clicked into this to read the differences and initial exp…

Well, I can share my experience from a few days ago. Gave the same task (a major refactor) to both Claude and Codex.

Codex finished in 5 minutes, Claude was still spinning after 20 minutes. Also it used up all my usage, about twice over (the 5-hour window rolled over in the middle of the task, so the usage for one task added up to 192%). Codex usage was 9%. So, 21x difference there, lol

They're saying there's bugs lately with how usage is being measured, but usage being buggy isn't exactly more encouraging...

So I was on task #4 with Codex while Claude was still spinning on #1.

I didn't like the results Codex gave me though. It has the habit of doing "technically what you asked, but not what a normal human would have wanted."

So given "Claude is great but I can't actually use it much" and "Codex is cheap and fast but kinda sucks", the current optimum seems to be having Claude write detailed specs and delegate to Codex. (OpenAI isn't banning people for using 3rd party orchestration, so this would actually be a thing you could do without problems. Not the reverse though.)

Re: Claude Opus 4.7

#680

Post by gck1 »

cmrdporcupine wrote: post #27
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...

Yep, I'll wait for the GPT answer to this. If we're lucky OpenAI will release a new GPT 5.5 or whatever model in the next few days, just like the last round. I have been getting better results out of codex on and off for months. It's more "careful" and systematic in its thinking. It makes less "excuses" and leaves less race conditions and slop around. And the actual codex CLI tool is better written, less buggy and fa…

What bothers me with codex cli is that it feels like it should be more observable, more open and verbose about what the model is doing per step, being an open source product and OpenAI seemingly being actually open for once, but then it does a tool call - "Read $file" and I have no idea whether it read the entire file, or a specific chunk of it. Claude cli shows you everything model is doing unless it's in a subagent (which is why I never use subagents).
Post reply on HN