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

721–730 of 1001 posts

Re: Claude Opus 4.7

#721

Post by slashdave »

computomatic wrote: post #72
cupofjoakim wrote: post #20

> Opus 4.7 uses an updated tokenizer that improves how the model processes text. The tradeoff is that the same input can map to more tokens—roughly 1.0–1.35× depending on the content type. caveman[0] is becoming more relevant by the day. I already enjoy reading its output more than vanilla so suits me well. [0] https://github.com/JuliusBrussee/caveman/tree/main

I was doing some experiments with removing top 100-1000 most common English words from my prompts. My hypothesis was that common words are effectively noise to agents. Based on the first few trials I attempted, there was no discernible difference in output. Would love to compare results with caveman. Caveat: I didn’t do enough testing to find the edge cases (eg, negation).

> My hypothesis was that common words are effectively noise to agents

Umm... a few words can be combined in a rather large number of ways.

Punctuation is used a lot. Why not just remove all the periods and commas and see what happens? Probably not pretty

Re: Claude Opus 4.7

#722

Post by clbrmbr »

andrepd wrote: post #440
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.

CoT is basically bullshit, entirely confabulated and not related to any "thought process"...

But still CoT distillation WORKS. See the DeepSeek R1 paper.

Re: Claude Opus 4.7

#723

Post by slashdave »

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 interpreted their actions as providing time for vendors to protect themselves against the new model proactively, not to nerf the models themselves.

Although perhaps I am naive.

Re: Claude Opus 4.7

#724

Post by HauntingPin »

vintagedave wrote: post #228
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...

Same. I stopped my Pro subscription yesterday after entering the week with 70% of my tokens used by Monday morning (on light, small weekend projects, things I had worked on in the past and barely noticed a dent in usage.) Support was... unhelpful. It's been funny watching my own attitude to Anthropic change, from being an enthusiastic Claude user to pure frustration. But even that wasn't the trigger to leave, it was…

I've given up on Claude after seeing the response quality degrade so much over the past two weeks, and now this? I've unsubscribed. I don't know why people are still giving this company money.

Re: Claude Opus 4.7

#725

Post by sbinnee »

bayesnet wrote: post #276

This is a CC harness thing than a model thing but the "new" thinking messages ('hmm...', 'this one needs a moment...') are extraordinarily irritating. They're both entirely uninformative and strictly worse than a spinner. On my workflows CC often spends up to an hour thinking (which is fine if the result is good) and seeing these messages does not build confidence.

The comment section is already long, but I knew that I could found comments about "hmm" that I started noticing. Yes, it is so irritating to me too. Also, one additional thing I noticed was that verbose information has been more and more being obfuscated. I run CC with --verbose option for months, and I can see verbose mode is not verbose anymore. I wish I can do -vvv maximum verbose mode.

Re: Claude Opus 4.7

#726

Post by b--l »

batshit_beaver wrote: post #175
endymion-light wrote: post #16

I'm not sure how much I trust Anthropic recently. This coming right after a noticeable downgrade just makes me think Opus 4.7 is going to be the same Opus i was experiencing a few months ago rather than actual performance boost. Anthropic need to build back some trust and communicate throtelling/reasoning caps more clearly.

What I want to know is why my bedrock-backed Claude gets dumber along with commercial users. Surely they're not touching the bedrock model itself. Only thing I can think of is that updates to the harness are the main cause of performance degradation.

If we learned anything from the code leak is that they essentially do not know what is in the blackbox of the code for that 500k line mass. So that's plausible.

Re: Claude Opus 4.7

#727

Post by esjeon »

aurareturn wrote: post #38
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...

Funny because many people here were so confident that OpenAI is going to collapse because of how much compute they pre-ordered. But now it seems like it's a major strategic advantage. They're 2x'ing usage limits on Codex plans to steal CC customers and it seems to be working. I'm seeing a lot of goodwill for Codex and a ton of bad PR for CC. It seems like 90% of Claude's recent problems are strictly lack of compute r…

Funny because the general consensus is that everyone is burning money so fast that they would not be able to get it back from their AI business in the near future. OpenAI is simply the one with the most aggressive expenditure. Google has its own cash cows. Anthropic has been conservative all around.

Re: Claude Opus 4.7

#728

Post by oezi »

The tokenizer changes seem to indicate that 4.7 isn't just a checkpoint but rather a model trained mostly from scratch, right?

Re: Claude Opus 4.7

#729

Post by b--l »

nl wrote: post #712

First model to get 100% on my agentic benchmark: https://sql-benchmark.nicklothian.com/?highlight=anthropic_c...

grok-4.1-fast is the the number 2 model on this benchmark.

~~If you've used this model in real life to do any sort of programming, and have seen its output, you would know that there is something VERY wrong with your benchmark.~~

Edit: Oh sorry, I looked at the questions, I see this is also for SQL specifically. Interesting. Maybe they tuned that grok model for SQL. Cool site. I bookmarked it.

Re: Claude Opus 4.7

#730

Post by johanyc »

varispeed wrote: post #363
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...

How do you get codex to generate any code? I describe the problem and codex runs in circles basically: codex> I see the problem clearly. Let me create a plan so that I can implement it. The plan is X, Y, Z. Do you want me to implement this? me> Yes please, looks good. Go ahead! codex> Okay. Thank you for confirming. So I am going to implement X, Y, Z now. Shall I proceeed? me> Yes, proceed. codex> Okay. Implementing.…

Weird. I never had that issue when writing code.
Post reply on HN