alt Hacker NewsClaude Opus 4.7
Comments
I happy with my GLM 5.1 and MiniMax 2.7 subscription and my wallet is happy, too.
I am glad Anthropic is pushing the limits, that means cheap Chinese models will have reasons to get better, too.
Pretty bad. As nerfed 4.6
I've been using 4.6 in a long-term development project every day for weeks.
4.7 is a clusterf--k and train wreck.
What’s the default context window? Seems extremely short.
while it seems even with 4.7 we will never see the quality of early 4.6 days, some dude is posting 'agi arrived!!!' on instagram and linkedIn.
Regardless of the model quality improvement, the corporate damage was done by not only ignoring the Opus quality degradation but gaslighting users into thinking they aren’t using it right.
I switched to Codex 5.4 xhigh fast and found it to be as good as the old Claude. So I’ll keep using that as my daily driver and only assess 4.7 on my personal projects when I have time.
Well this explains the outages over the last few days
Uh oh:
> The new /ultrareview slash command produces a dedicated review session that reads through changes and flags bugs and design issues that a careful reviewer would catch. We’re giving Pro and Max Claude Code users three free ultrareviews to try it out.
As Anthropic keeps pushing the pricing envelope wider it makes room for differentiation, which is good. But I wish oAI would get a capable agentic model out the door that pushes back on pricing.
Ps I know that Anthropic underbought compute and so we are facing at least a year of this differentiated pricing from them, but still..ouch
So Mythos.
I get a little sad with every new Claude release. Sonnet 4.5 is my favorite and each new model means it's one step closer to being retired. Nothing else replaces it for me
four prompts with opus 4.6 today is equivalent to 30 or 40 two months ago. infernal downgrade in my case.
show us the benchmarks with "adaptive thinking" turned on
Will it be like the usual: let it work great for 2 weeks, nerf it after?
This is the first new model from Anthropic in a while that I'm not super enthused about. Not because of the model, I literally haven't opened the page about it, I can already guess what it says ("Bigger, better, faster, stronger"), but because of the company.
I have enjoyed using Claude Code quite a bit in the past but that has been waning as of late and the constant reports of nerfed models coupled with Anthropic not being forthcoming about what usage is allowed on subscriptions [0] really leaves a bad taste in my mouth. I'll probably give them another month but I'm going to start looking into alternatives, even PayG alternatives.
[0] Please don't @ me, I've read every comment about how it _is clear_ as a response to other similar comments I've made. Every. Single. One. of those comments is wrong or completely misses the point. To head those off let me be clear:
Anthropic does not at all make clear what types of `claude -p` or AgentSDK usage is allowed to be used with your subscription. That's all I care about. What am I allowed to use on my subscription. The docs are confusing, their public-facing people give contradictory information, and people commenting state, with complete confidence, completely wrong things.
I greatly dislike the Chilling Effect I feel when using something I'm paying quite a bit (for me) of money for. I don't like the constant state of unease and being unsure if something might be crossing the line. There are ideas/side-projects I'm interested in pursuing but don't because I don't want my account banned for crossing a line I didn't know existed. Especially since there appears to be zero recourse if that happens.
I want to be crystal clear: I am not saying the subscription should be a free-for-all, "do whatever you want", I want clear lines drawn. I increasingly feeling like I'm not going to get this and so while historically I've prefered Claude over ChatGPT, I'm considering going to Codex (or more likely, OpenCode) due to fewer restrictions and clearer rules on what's is and is not allowed. I'd also be ok with kind of warning so that it's not all or nothing. I greatly appreciate what Anthropic did (finally) w.r.t. OpenClaw (which I don't use) and the balance they struck there. I just wish they'd take that further.
"Agentic Coding/Terminal/Search/Analysis/Etc"...
False: Anthropic products cannot be used with agents.
As one of the seemingly few people in this comments section who don't use it for coding, it seems far far more substantial and able to produce insights in written conversation than opus 4.6 for me
Getting a little suspicious that we might not actually get AGI.
> 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 requests that indicate prohibited or high-risk cybersecurity uses.
Ah f... you!
is this just mythos flex?
Is that time to turning back from Codex to Claude Code?
its a pretty good coding model - using it in cursor now.
It’s funny, a few months ago I would have been pretty excited about this. But I honestly don’t really care because I can’t trust Anthropic to not play games with this over the next month post release.
I just flat out don’t trust them. They’ve shown more than enough that they change things without telling users.
Really disappointed with Anthropic recently, burned through 2 max plans and extra usage past 10 days, getting limited almost 1h in a 5h session. Reading about the extra "safe guards" might be the nail on the coffin.
amazing speed...
yay! lobotomized mythos is out
Opus 4.7 would come out the day before my paid plan ends.
What a joke Opus 4.7 at max is.
I gave it an agentic software project to critically review.
It claimed gemini-3.1-pro-preview is wrong model name, the current is 2.5. I said it's a claim not verified.
It offered to create a memory. I said it should have a better procedure, to avoid poisoning the process with unverified claims, since memories will most likely be ignored by it.
It agreed. It said it doesn't have another procedure, and it then discovered three more poisonous items in the critical review.
I said that this is a fabrication defect, it should not have been in production at all as a model.
It agreed, it said it can help but I would need to verify its work. I said it's footing me with the bill and the audit.
We amicably parted ways.
I would have accepted a caveman-style vocabulary but not a lobotomized model.
I'm looking forward to LobotoClaw. Not really.
just started using codex. claude is just marketing machine and benchmaxxing and only if you pay gazillion and show your ID you can use their dangerous model.
This new one seems even pushier to shove me on the shortest-path solution
We've all been complaining about Opus 4.6 for weeks and now there's a new model. Did they intentionally gimp 4.6 so they can advertise how much better 4.7 is?
This is the 7th advert on the front page right now. It's ridiculous
Might be sticking with 4.6 it's only been 20 minutes of using 4.7 and there are annoyances I didn't face with 4.6 what the heck. Huge downgrade on MRCR too....
256K:
- Opus 4.6: 91.9% - Opus 4.7: 59.2%
1M:
- Opus 4.6: 78.3% - Opus 4.7: 32.2%
Codex release coming today: https://x.com/thsottiaux/status/2044803491332526287
Here’s the problem. The distribution of query difficulty / task complexity is probably heavily right-skewed which drives up the average cost dramatically. The logical thing for anthropic to do, in order to keep costs under control, is to throttle high-cost queries. Claude can only approximate the true token cost of a given query prior to execution. That means anything near the top percentile will need to get throttled as well.
By definition this means that you’re going to get subpar results for difficult queries. Anything too complicated will get a lightweight model response to save on capacity. Or an outright refusal which is also becoming more common.
New models are meaningless in this context because by definition the most impressive examples from the marketing material will not be consistently reproducible by users. The more users who try to get these fantastically complex outputs the more those outputs get throttled.
Reminder that 4.7 may seem like a huge upgrade to 4.6 because they nerfed the F out of 4.6 ahead of this launch so 4.7 would seem like a remarkable improvement...
All fine, where is pelican on bicycle?
> First, Opus 4.7 uses an updated tokenizer that improves how the model processes text
wow can I see it and run it locally please? Making API calls to check token counts is retarded.
Excited to start using this!
Introducing a new upgraded slot machine named "Claude Opus" in the Anthropic casino.
You are in for a treat this time: It is the same price as the last one [0] (if you are using the API.)
But it is slightly less capable than the other slot machine named 'Mythos' the one which everyone wants to play around with. [1]
Sigh here we go again, model release day is always the worst day of the quarter for me. I always get a lovely anxiety attack and have to avoid all parts of the internet for a few days :/
even sonnet right now has degraded for me to the point of like ChatGPT 3.5 back then. took ~5 hours on getting a playwright e2e test fixed that waited on a wrong css selector. literlly, dumb as fuck. and it had been better than opus for the last week or so still... did roughly comparable work for the last 2 weeks and it all went increasingly worse - taking more and more thinking tokens circling around nonsense and just not doing 1 line changes that a junior dev would see on the spot. Too used to vibing now to do it by hand (yeah i know) so I kept watching and meanwhile discovered that codex just fleshed out a nontrivial app with correct financial data flows in the same time without any fuzz. I really don't get why antrhopic is dropping their edge so hard now recently, in my head they might aim for increasing hype leading to the IPO, not disappointment crashes from their power user base.
"Error: claude-opus-4-6[1m] is temporarily unavailable".
I'm an Opus fanboy, but this is literally the worst coding model I have used in 6 months. Its completely unusable and borderline dangerous. It appears to think less than haiku, will take any sort of absurd shortcut to achieve its goal, refuses to do any reasoning. I was back on 4.6 within 2 hours.
Did Anthropic just give up their entire momentum on this garbage in an effort to increase profitability?