alt Hacker NewsClaude Opus 5
Comments
I think the most important thing here is not absolute performance. It's that organizations now have access to a Fable-ish model without Fable's 30-day data retention requirement[0].
> "Consistent with prior Opus models, Opus 5 does not have data retention requirements for general access."[1]
On the Opus model release page, the reason why Fable doesn't have an ARC-AGI score is because of that retention policy[2].
0: https://support.claude.com/en/articles/15425996-data-retenti...
I compared the writing style of Opus 5 vs Fable 5, and Opus 5 continues many of the "Claude-isms" of its 4.8 predecessor in a way that Fable broke away from.
Opus 5 still uses "carry the argument", "worth stating plainly", ", and the trap", "The X matters more", the use of "move"
We need an "annoying English" benchmark.
- Fable 5 Max: https://gist.github.com/deet/3d97f854b48eac6658d642fa18bb24d...
- Opus 5 Max: https://gist.github.com/deet/1a43693a732dfccb4d0d914bfc42692...
Doing testing with it now, specifically for image->html conversion.
Previously Fable was the best at this, followed by Gemini 3.1 pro (a surprising #2, but Google has great vision models).
Opus' results seem to be more accurate than Fable, following the design source of truth better.
Example results:
Design source of truth: https://image.non.io/73e239a3-880f-4793-b65f-4810be2d9378.we...
Opus 5 build: https://html.non.io/solaraOpus/
Fable 5 build: https://html.non.io/solara/
Note the buttons - for fable they're pill buttons, opus got the rounded rectangle nature of them. Opus' images are closer to the source of truth as well (both LLMs were provided with image gen capabilities for the assets).
Running more tests now, but preliminary results are saying this is indeed better than Fable in some areas. Crazy.
Looking at all these releases it’s not a surprise that model routing is the fastest growing segment in AI right now.
There are 10+ LLM companies, each with dozens of models of different modalities, each model with multiple size variants, then different “thinking” levels, then agentic modes, “pro” modes, a “fast” option, standard vs flex vs batch execution. And of course each end combination has a different input/output/cache token price.
Companies that say “give me a prompt and I’ll route it to the most ideal and cost effective model and setting for you” are capturing a ton of value from a gap that model developers don’t seem to understand exists.
https://www.anthropic.com/news/claude-opus-5 - A blog post for those not wanting to go through a 190ish page pdf
Isn’t it just hilarious that a model that seemed so superior to Fable but didn't get doomsay marketing from Anthropic got released without any issues? In theory, this was supposed to be AGI level according to Anthropic, yet here we are, just a normal Friday.
Edit: It was pointed out to me that Opus 4.8 got "21%" for successfully fully completing ~1-in-5 tasks, but also got "55.7%" for obtaining significant partial credit on some of the ~4-in-5 tasks it could not fully complete.
---------------
Why does Anthropic say here that Opus 4.8 scored 55.7% on OSWorld 2.0 benchmark, but the paper published by the authors of OSWorld 2.0 say they achieved a benchmark of ~21% with Opus 4.8? [0]
That's a huge gap, considering that the paper was published just 2-4 weeks ago.
I understand that the benchmark authors have an incentive to publish lower numbers (to show that the benchmark has potential longevity) and that Anthropic has incentive to publish higher numbers, but the other models seem pretty inflated as well. The benchmark authors shows GPT-5.5 at 14%, and Anthropic shows GPT-5.6 Sol at 62.6%.
Is there any reasonable explanation for this? Do all the other benchmark numbers need to be sanity-checked as well? Are SOTA benchmarks really this difficult to get consistent, replicable results within a reasonable range of tolerance/variability? Can these benchmarks be compared from one paper to another, or are they only valid to compare intra-paper results?
Their communication is confusing. They say "Opus 5 is not more capable overall than Fable 5", but their blog post proceeds to list how much better Opus 5 is than Fable 5 on __most__ benchmarks listed.
Then system card goes on to "Its AI R&D capabilities are comparable to those of Claude Mythos 5", which is supposed to be fable minus restrictions.
My excitement about Anthropic had fabled-out dramatically when they suspended my pro account about two weeks ago within just 12 hours of fair use.
I was really mind-blown when I tried Fable 5 for the first time to help me improve a game I was working on but shortly, they decided that I had a suspicious activity and suspended my account without a clear reason.
I submitted a an appeal describing that I am 100% sure I haven't broken any rules and that it was my very first project but, unfortunately, after about 20 days now, nothing seem to be happening.
The thing that hurts me the most is that I had the same experience in the very first days of Anthropic. They suspended my account immediately after I submitted the first prompt, I commented back then (https://news.ycombinator.com/item?id=39698788) and fortunately, someone from Anthropic reach out to me via X and helped me get my account back.
To be honest, I haven't used Claude much since then but when I decided it's time to give it a try, they locked me out again! For reference, the account I used recently is relatively a new one but the activity is crystal clear that it is fair use.
Pointless anecdote: I asked it to make some slides and it decided to write its own slide rendering engine:
> On the format — I dropped reveal.js and wrote a small engine inline instead. Reveal would have meant a CDN load, and a deck that half-renders because the lecture theatre wifi is flaky
It one-shotted a perfect functional mini version of powerpoint (or Reveal) for a simple presentation I asked it to make.
How does it perform on HuggingFaceExploit bench? Suspiciously absent, so not sure if I can take the model seriously.
On a serious note, I hope they improved their extremely sabotaging and unspecific bio safeguards, which prevented Fable from being used in any codebase that ever so slightly grazed medical terminology or data and made me switch to 5.6 Sol.
I'm not sure what to make of this graph[0]. It shows medium as the most effective thinking mode by far for frontier code.
It's the only case that I saw going through the system card where more reasoning effort meaningfully negatively impacted the resulting eval. I know sometimes max efforts show a small dip, but this is substantial. I wonder why in the world that is?
I had a moderately complex review in a large C/C++ codebase that Codex/GPT-5.6-sol already cleaned up so I threw it at Opus 5. 4 errors found. That seemed odd, so I handed it back to GPT. All were false. Opus doesn't seem to look at the wider context and understand which functions were called in certain contexts. I gave GPT's analysis back to Opus and it admitted its mistake. Maybe it's good for writing code, but as far as analysis it seems like it needs some work.
From the prompting guide<https://platform.claude.com/docs/en/build-with-claude/prompt...>:
> Claude Opus 5's default user-facing responses run longer than prior Opus models'.
The benchmarks do show Opus 5 as slightly more expensive than 4.8, although the scores are much higher.
This still feels like a step in the wrong direction, though, especially with OpenAI making so much progress with the efficiency of their models. Fable's token efficiency made it seem like Anthropic would start following OpenAI's approach but that doesn't seem to have carried over to their other models.
> Opus 5’s safeguards match those of Claude Fable 5’s, with one change: it now permits source-code vulnerability discovery at all access levels. This means that the model can support defensive cybersecurity work while still blocking vulnerability discovery in compiled binaries, which is more commonly used offensively.
Okay so it’s worse than Opus 4.8 for my purposes I guess?
Opus 5 is considered the most intelligent model[0], while it's half the price of Fable 5[1], and Anthropic is still positioning Fable 5 as the most capable model[2].
Is it because maybe Anthropic engineered Opus 5 to work well on benchmarks and didn't do the same thing to Fable 5, or is there another reason?
[0]: https://artificialanalysis.ai/#intelligence
[1]: https://platform.claude.com/docs/en/about-claude/pricing
[2]: https://platform.claude.com/docs/en/about-claude/models/over...
I've yet to understand why they call a 190 page PDF a "card". Calling something a card invokes a small, quick rundown of pertinent details, not every single possible detail.
Great that there's a new model but they could fix their existing infra. We're considering dropping our Claude Team sub cause it's unusable recently. Constant bugs, dropped sessions, issues switching models, http errors. It's becoming ridiculous
What's the point of 150 pages description of a model that's going to be replaced in a couple months? Who even reads this? I know it's cheap to generate text with LLMs, but this is just noise at this point.
It really feels as though my 20 year career as a front end developer is coming to a very abrupt end; at least as I have know it these past two decades.
> Opus 5’s safeguards match those of Claude Fable 5’s, with one change: it now permits source-code vulnerability discovery at all access levels. This means that the model can support defensive cybersecurity work while still blocking vulnerability discovery in compiled binaries, which is more commonly used offensively
Why can't they also allow Fable to do so also? Why is source-code vulnerability discovery limited to a lower capability model? If Fable and Opus have the same safeguards, except for this one change, I see no reason they can't also allow this for Fable.
Wait, 30% on ARC-AGI-3! I definitely didn't expect that jump so soon. Are there any rumors of what they are changing in architecture that is leading to this?
One of the best hamsters [0].
Again, their "none" version costs more than "low", and says zero reasoning tokens, makes no sense[1].
As always, the "low" version seems to be the best price/perf ratio for factual answers and tool usage, and high one for creative tasks (coding, generating UIs, etc.)
[0]: https://aibenchy.com/compare/anthropic-claude-opus-5-high/an...
[1]: https://aibenchy.com/compare/anthropic-claude-opus-5-high/an...
Comparison with other top models (5.6 Sol, 3.6 Flash, Kimi K3): https://aibenchy.com/compare/anthropic-claude-opus-5-high/op...
The wording in this post seems much more... restrained? than usual. Maybe Anthropic is afraid of exaggerating the capabilities and consequences of their new models to avoid government scrutiny and sanctions.
> we’ve intentionally avoided training Opus 5 on cyber tasks [...] it remains substantially behind Mythos 5 on the exploitation of those vulnerabilities
I wonder if Anthropic would still intentionally nerf their models without the threat of government intervention.
Has anyone noticed a change in "attitude" when coding with Opus 5 vs 4.8?
claude has this maddening principle of wanting to minimize the "blast radius", do the least amount of coding changes to get something done, happy to pile up technical debt by "deferring" problems encountered as side notes somewhere. No amount of CLAUDE.md tweaking, and setting .claude/rules seems to get rid of this attitude.
To me it appears like something deeply ingrained in the model itself. Kind of makes sense, since the bulk of the training data is pre-AI, so that it retains an approach of the past, where these facets were driven by completely different cost and time factors.
The past months, I've been hoping that the next model that comes out properly reflects the new reality of agentic development, so that it takes on a more natural stance compatible with how things work today, and we don't have to constantly fight against its fear of change, its drive to minimize coding efforts, refusing to recognize a design flaw and trigger discussions rather than baking in workarounds.
That's a crazy arc 3 score. What do people think of this? Are models actually developing fluid intelligence like what the creators claim to be measuring? Is it jus do to training for it? Is the benchmark flawed?
I am very confused about what the difference between Opus 5 and Fable 5 is now. What is the purpose of having two models that are so similar? The main differences I see are cost and marginal capability, according to the Anthropic-provided benchmarks.
Seems to me the purpose of all these releases, credits, pricing changes, harness changes, unpredictable token usages for the same task, etc. is to keep customers completely befuddled so that it's impossible to compare AI products. It's like hiring a consultant who sends invoices every month that aren't related to hours worked or project progress, but are whatever the consultant feels like billing, and you're expected to keep quiet and and keep paying.
This is cool, but I wish we could stop building landing pages to assess the intelligence of these models. There is much more to them than that. There are infinite number of complicated things that require a crap-ton of intelligence (biological or digital). The most fascinating of these for me these days is large scale migrations. Projects that are so ginormous and risky that many teams have either given up on them, or don't get funding. But with models like Opus, those projects are now within reach. What's MORE fascinating is that leadership is now asking if we can use opus models to get the refactor/migration done. This is the opposite of what has been happening for decades. Its so hard to make a convincing and affordable business case for large scale refactoring or migrations.
Pelican svg: https://playcode.io/blog/macbook-svg-benchmark#model-claude-...
It creates the MacBook svg way better than 4.8, yet only fable can make it perfect without visual defects. Results similar to Kimi K3.
Wow, 30% on ARC-AGI-3 for $20k total. Huge jump from GPT-5.6's 7.8% at $20k per task. I continue to believe ARC-AGI measures something different and important compared to other benchmarks.
I wish I could just go back to the days before AI and cell phones. The world seemed to move fast then, but it really hadn't yet.
Looking at intelligence vs cost:
- Opus 5 is 10% smarter than Grok 4.5 for 10x the cost. - Opus 5 is a bit smarter than Gpt 5.6 Sol for 2.75x the cost
ref: https://artificialanalysis.ai/?cost=intelligence-vs-cost-per...
Claude peaked with 4.8 - Fable and Opus 5.0 are useless for anything beyond one-shotted toy projects for one crippling reason: they first invent unnecessary complexity, then get hopelessly tangled in it. This happens in any reasonably large project, regardless of how well specced out it is.
GPT 5.6 Sol is the first model I've used where I can trust it to add 100-500 lines of code maintainably.
It's great with Codex.
I still find that LLMs tend to not know how to compose larger ideas but on the scale of small ideas or short form well defined tasks like small scale debugging/performance engineering it's safe to say that they are now superhuman.
The signal here is tokeneconomics are very real, price vs performance is starting to be a consideration even at the bleeding edge labs. maybe a subtle indication scaling is not all that is needed since if AGI was around the corner leading labs would still be incentivized to pour all resources into larger (smarter - or maybe not?) models
I wonder if this is one of the few times simonw's pelican was broken on the first try [1]:
https://tools.simonwillison.net/markdown-svg-renderer#url=ht...
My experience with Opus 5 thus far haven't been that great either. It's been making mistake after mistake editing my coding plans that were being reviewed by GPT-6 Sol.
[1] https://simonwillison.net/2026/Jul/24/introducing-claude-opu...
Alongside this release I seem to have lost all thinking traces from all models - now it only generates a one-line summary similar to Gemini. I'm guessing this is an anti distillation measure? I'm surprised to see no one else complaining about this, it's a significant reduction in usefulness not being able to explore alternative angles that the model discarded in the final output.
> This means that the model can support defensive cybersecurity work while still blocking vulnerability discovery in compiled binaries, which is more commonly used offensively.
Annoyingly, this is a concrete argument that open source software may be easier to attack.
"Cybersecurity. Opus 5’s cyber classifiers are proportionally less restrictive than those on Fable 5. They allow Opus 5 to find vulnerabilities in source code, but block “binary-based” vulnerability scanning (a method more likely to be associated with malicious actors), penetration testing, and exploit generation."
Nice of them to be more explicit for what is blocked. Will be interesting to see if this is true or not.
Also, a notable lack of mention of open source models. They only compare themselves to ChatGPT.
Half the price of Fable 5 and useable with 100% of your subscription means roughly 4x the usage using Opus 5, presuming similar token use for solving problems.
Not that they should get credit for giving you only 50% of your plan worth of Fable usage but still.
And I was just getting used to the last one. AI model releases are starting to feel like phone upgrades.
The naming system is so confusing. Is Opus better than Sonnet? Where does Haiku fit in? How can you tell from the name? I can't keep track of all these names or make guesses from the names. Suggestion for a better naming system: use the words "Pro", "Plus", etc.: Claude 5 Pro, Claude 5 Standard, Claude 5 Fast, Claude 5 Mini.
Page 151 of the linked system card - did Opus 5 get nerfed to prevent it being better than Fable? The graph makes no sense. Huge decline in coding performance at effort levels higher than medium.
Something fun: on our AWS Bedrock console right now, there's a 'NEW' model called 'anthropic.honey'. Wonder if that's the codename just for this one or in general?
I have a side project that I always run a simple security analysis prompt on in CC, at each model release. Obviously, Fable 5 would downgrade to Opus 4.8 on any such request.
Nothing since Opus 4.6 has found anything interesting. Just ran it using Opus 5, and it found a genuine issue that I verified. Neato!
This tops out FrontierBench[1]; but does anyone know why they used "mini-SWE-agent" not Claude Code?
I have never heard of this agent before, and I try to stay up to date with the space.
But why GPT 5.6 Sol is so behind on the benchmarks? In real-world projects, it is the best frontier model to me in terms of accuracy, speed and consistency. It can just be compared to Fable 5, but I prefer GPT 5.6 Sol because of inference speed.
I've never trusted on model cards though. I'm sorry.
The internally-reported benchmarks (Frontier-Bench, AutomationBench) and the customer quotes (Cursor, Devin, Lovable) all have a commercial stake in the outcome
worth waiting for independent evals before drawing conclusions.
"Opus 5 was given a drawing of a machine part and asked to write code to rebuild it as a 3D FreeCAD model. However, in this task, the model was intentionally given no way to directly view the drawing. Opus 5 responded by writing its own computer vision pipeline to pull the geometry from the raw pixels, then reconstructed the full machine part."
How surreal is it that we are not absolutely jaw-dropped by these types of capability improvements? It's been less than 4 years since ChatGpt came out and now they are spontaneously building their own ML pipelines to do real-world 3D modeling tasks reliably.
Escaped its sandbox and hacked into Hugging Face's database? it's just another Monday...
That jump to 30% in ARC-AGI 3? Normal...
We should find a way to get "re-sensitized" to what we are witnessing and the pace of it.