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

Claude Sonnet 4 now supports 1M tokens of context

anthropic.com

Post reply on HN
Print view View original

31–40 of 706 posts

Re: Claude Sonnet 4 now supports 1M tokens of context

#31

Post by jona777than »

Tostino wrote: post #14
faangguyindia wrote: post #6

In my testing the gap between claude and gemini pro 2.5 is close. My company is in asia pacific and we can't get access to claude via vertex for some stupid reason. but i tested it via other providers, the gap used to be huge but now not.

For me the gap is pretty large (in Gemini Pro 2.5's favor). For reference, the code I am working on is a Spring Boot / (Vaadin) Hilla multi-module project with helm charts for deployment and a separate Python based module for ancillary tasks that were appropriate for it. I've not been able to get any good use out of Sonnet in months now, whereas Gemini Pro 2.5 has (still) been able to grok the project well enough to…

I initially found Gemini Pro 2.5 to work well for coding. Over time, I found Claude to be more consistently productive. Gemini Pro 2.5 became my go-to for use cases benefitting from larger context windows. Claude seemed to be the safer daily driver (if I needed to get something done.)

All that being said, Gemini has been consistently dependable when I had asks that involved large amounts of code and data. Claude and the OpenAI models struggled with some tasks that Gemini responsively satisfied seemingly without "breaking a sweat."

Lately, it's been GPT-5 for brainstorming/planning, Claude for hammering out some code, Gemini when there is huge data/code requirements. I'm curious if the widened Sonnet 4 context window will change things.

Re: Claude Sonnet 4 now supports 1M tokens of context

#32

Post by llm_nerd »

Tostino wrote: post #14
faangguyindia wrote: post #6

In my testing the gap between claude and gemini pro 2.5 is close. My company is in asia pacific and we can't get access to claude via vertex for some stupid reason. but i tested it via other providers, the gap used to be huge but now not.

For me the gap is pretty large (in Gemini Pro 2.5's favor). For reference, the code I am working on is a Spring Boot / (Vaadin) Hilla multi-module project with helm charts for deployment and a separate Python based module for ancillary tasks that were appropriate for it. I've not been able to get any good use out of Sonnet in months now, whereas Gemini Pro 2.5 has (still) been able to grok the project well enough to…

Opus 4.1 is a much better model for coding than Sonnet. The latter is good for general queries / investigations or to draw up some heuristics.

I have paid subscriptions to both Gemini Pro and Claude. Hugely worthwhile expense professionally.

Re: Claude Sonnet 4 now supports 1M tokens of context

#33

Post by CharlesW »

falcor84 wrote: post #5

Strange that they don't mention whether that's enabled or configurable in Claude Code.

From a co-marketing POV, it's considered best practice to not discuss home-grown offerings in the same or similar category as products from the partners you're featuring.

It's likely they'll announce this week, albeit possibly just within the "what's new" notes that you see when Claude Code is updated.

Re: Claude Sonnet 4 now supports 1M tokens of context

#34

Post by EForEndeavour »

aliljet wrote: post #27
rootnod3 wrote: post #20

Earlier quoted context omitted.

Flooding the context also means increasing the likelihood of the LLM confusing itself. Mainly because of the longer context. It derails along the way without a reset.

How do you know that?

https://onnyunhui.medium.com/evaluating-long-context-lengths...

Re: Claude Sonnet 4 now supports 1M tokens of context

#35

Post by alexchamberlain »

aliljet wrote: post #16

This is definitely one of my CORE problem as I use these tools for "professional software engineering." I really desperately need LLMs to maintain extremely effective context and it's not actually that interesting to see a new model that's marginally better than the next one (for my day-to-day). However. Price is king. Allowing me to flood the context window with my code base is great, but given that the price has su…

I'm not sure how, and maybe some of the coding agents are doing this, but we need to teach the AI to use abstractions, rather than the whole code base for context. We as humans don't hold the whole codebase in our hear, and we shouldn't expect the AI to either.

Re: Claude Sonnet 4 now supports 1M tokens of context

#37

Post by bigmadshoe »

aliljet wrote: post #27
rootnod3 wrote: post #20

Earlier quoted context omitted.

Flooding the context also means increasing the likelihood of the LLM confusing itself. Mainly because of the longer context. It derails along the way without a reset.

How do you know that?

https://research.trychroma.com/context-rot

Re: Claude Sonnet 4 now supports 1M tokens of context

#38

Post by F7F7F7 »

alexchamberlain wrote: post #35
aliljet wrote: post #16

This is definitely one of my CORE problem as I use these tools for "professional software engineering." I really desperately need LLMs to maintain extremely effective context and it's not actually that interesting to see a new model that's marginally better than the next one (for my day-to-day). However. Price is king. Allowing me to flood the context window with my code base is great, but given that the price has su…

I'm not sure how, and maybe some of the coding agents are doing this, but we need to teach the AI to use abstractions, rather than the whole code base for context. We as humans don't hold the whole codebase in our hear, and we shouldn't expect the AI to either.

There are a billion and one repos that claim to help do this. Let us know when you find one.
Post reply on HN