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

Claude Sonnet 4.6

anthropic.com

Post reply on HN
Print view View original

281–290 of 1001 posts

Re: Claude Sonnet 4.6

#281

Post by holoduke »

fullstackchris wrote: post #199
andsoitis wrote: post #77

I’m voting with my dollars by having cancelled my ChatGPT subscription and instead subscribing to Claude. Google needs stiff competition and OpenAI isn’t the camp I’m willing to trust. Neither is Grok. I’m glad Anthropic’s work is at the forefront and they appear, at least in my estimation, to have the strongest ethics.

idk, codex 5.3 frankly kicks opus 4.6 ass IMO... opus i can use for about 30 min - codex i can run almost without any break

What about the client ? I find the Claude cliënt better in planning, making the right decision steps etc. it seems that a lot of work is also in the cli tool itself. Specially in feedback loop processing (reading logs. Browsers. Consoles etc)

Re: Claude Sonnet 4.6

#282

Post by data-ottawa »

andrewchilds wrote: post #244

Many people have reported Opus 4.6 is a step back from Opus 4.5 - that 4.6 is consuming 5-10x as many tokens as 4.5 to accomplish the same task: https://github.com/anthropics/claude-code/issues/23706 I haven't seen a response from the Anthropic team about it. I can't help but look at Sonnet 4.6 in the same light, and want to stick with 4.5 across the board until this issue is acknowledged and resolved.

I think this depends on what reasoning level your Claude Code is set to.

Go to /models, select opus, and the dim text at the bottom will tell you the reasoning level.

High reasoning is a big difference versus 4.5. 4.6 high uses a lot of tokens for even small tasks, and if you have a large codebase it will fill almost all context then compact often.

Re: Claude Sonnet 4.6

#283

Post by skissane »

imiric wrote: post #236
spondyl wrote: post #183

Earlier quoted context omitted.

If you read the resignation letter, they would appear to be so cryptic as to not be real warnings at all and perhaps instead the writings of someone exercising their options to go and make poems

[flagged]

> Let's ignore the words of a safety researcher from one of the most prominent companies in the industry

I think "safety research" has a tendency to attract doomers. So when one of them quits while preaching doom, they are behaving par for the course. There's little new information in someone doing something that fits their type.

Re: Claude Sonnet 4.6

#284

Post by reed1234 »

reed1234 wrote: post #261
andrewchilds wrote: post #244

Many people have reported Opus 4.6 is a step back from Opus 4.5 - that 4.6 is consuming 5-10x as many tokens as 4.5 to accomplish the same task: https://github.com/anthropics/claude-code/issues/23706 I haven't seen a response from the Anthropic team about it. I can't help but look at Sonnet 4.6 in the same light, and want to stick with 4.5 across the board until this issue is acknowledged and resolved.

not in my experience

"Opus 4.6 often thinks more deeply and more carefully revisits its reasoning before settling on an answer. This produces better results on harder problems, but can add cost and latency on simpler ones. If you’re finding that the model is overthinking on a given task, we recommend dialing effort down from its default setting (high) to medium."[1]

I doubt it is a conspiracy.

[1] https://www.anthropic.com/news/claude-opus-4-6

Re: Claude Sonnet 4.6

#285

Post by Foobar8568 »

andrewchilds wrote: post #244

Many people have reported Opus 4.6 is a step back from Opus 4.5 - that 4.6 is consuming 5-10x as many tokens as 4.5 to accomplish the same task: https://github.com/anthropics/claude-code/issues/23706 I haven't seen a response from the Anthropic team about it. I can't help but look at Sonnet 4.6 in the same light, and want to stick with 4.5 across the board until this issue is acknowledged and resolved.

It goes into plan mode and/or heavy multiple agent for any reasons, and hundred thousands of tokens are used within a few minutes.

Re: Claude Sonnet 4.6

#286

Post by bsamuels »

grav wrote: post #278
andrewchilds wrote: post #244

Many people have reported Opus 4.6 is a step back from Opus 4.5 - that 4.6 is consuming 5-10x as many tokens as 4.5 to accomplish the same task: https://github.com/anthropics/claude-code/issues/23706 I haven't seen a response from the Anthropic team about it. I can't help but look at Sonnet 4.6 in the same light, and want to stick with 4.5 across the board until this issue is acknowledged and resolved.

I fail to understand how two LLMs would be "consuming" a different amount of tokens given the same input? Does it refer to the number of output tokens? Or is it in the context of some "agentic loop" (eg Claude Code)?

thinking tokens, output tokens, etc. Being more clever about file reads/tool calling.

Re: Claude Sonnet 4.6

#287

Post by dalmo3 »

imiric wrote: post #236
spondyl wrote: post #183

Earlier quoted context omitted.

If you read the resignation letter, they would appear to be so cryptic as to not be real warnings at all and perhaps instead the writings of someone exercising their options to go and make poems

[flagged]

Weak appeal to fiction fallacy.

Also, trajectory of celestial bodies can be predicted with a somewhat decent level of accuracy. Pretending societal changes can be equally predicted is borderline bad faith.

Re: Claude Sonnet 4.6

#289

Post by jcims »

grav wrote: post #278
andrewchilds wrote: post #244

Many people have reported Opus 4.6 is a step back from Opus 4.5 - that 4.6 is consuming 5-10x as many tokens as 4.5 to accomplish the same task: https://github.com/anthropics/claude-code/issues/23706 I haven't seen a response from the Anthropic team about it. I can't help but look at Sonnet 4.6 in the same light, and want to stick with 4.5 across the board until this issue is acknowledged and resolved.

I fail to understand how two LLMs would be "consuming" a different amount of tokens given the same input? Does it refer to the number of output tokens? Or is it in the context of some "agentic loop" (eg Claude Code)?

One very specific and limited example, when asked to build something 4.6 seems to do more web searches in the domain to gather latest best practices for various components/features before planning/implementing.

Re: Claude Sonnet 4.6

#290

Post by moritzwarhier »

Certhas wrote: post #260
moritzwarhier wrote: post #147

Earlier quoted context omitted.

Deceptive is such an unpleasant word. But I agree. Going back a decade: when your loss function is "survive Tetris as long as you can", it's objectively and honestly the best strategy to press PAUSE/START. When your loss function is "give as many correct and satisfying answers as you can", and then humans try to constrain it depending on the model's environment, I wonder what these humans think the specification for…

Correct and satisfying answers is not the loss function of LLMs. It's next token prediction first.

Thanks for correcting; I know that "loss function" is not a good term when it comes to transformer models.

Since I've forgotten every sliver I ever knew about artificial neural networks and related basics, gradient descent, even linear algebra... what's a thorough definition of "next token prediction" though?

The definition of the token space and the probabilities that determine the next token, layers, weights, feedback (or -forward?), I didn't mention any of these terms because I'm unable to define them properly.

I was using the term "loss function" specifically because I was thinking about post-training and reinforcement learning. But to be honest, a less technical term would have been better.

I just meant the general idea of reward or "punishment" considering the idea of an AI black box.

Post reply on HN