Live data from Hacker News

Claude Opus 4.6

anthropic.com

61–70 of 1001 posts

Re: Claude Opus 4.6

#61
Impressive that they publish and acknowledge the (tiny, but existent) drop in performance on SWE-Bench Verified between Opus 4.5 to 4.6. Obviously such a small drop in a single benchmark is not that meaningful, especially if it doesn't test the specific focus areas of this release (which seem to be focused around managing larger context).

But considering how SWE-Bench Verified seems to be the tech press' favourite benchmark to cite, it's surprising that they didn't try to confound the inevitable "Opus 4.6 Releases With Disappointing 0.1% DROP on SWE-Bench Verified" headlines.

Re: Claude Opus 4.6

#62
Claude Code release notes:

  > Version 2.1.32:
     • Claude Opus 4.6 is now available!
     • Added research preview agent teams feature for multi-agent collaboration (token-intensive feature, requires setting
     CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1)
     • Claude now automatically records and recalls memories as it works
     • Added "Summarize from here" to the message selector, allowing partial conversation summarization.
     • Skills defined in .claude/skills/ within additional directories (--add-dir) are now loaded automatically.
     • Fixed @ file completion showing incorrect relative paths when running from a subdirectory
     • Updated --resume to re-use --agent value specified in previous conversation by default.
     • Fixed: Bash tool no longer throws "Bad substitution" errors when heredocs contain JavaScript template literals like ${index + 1}, which
     previously interrupted tool execution
     • Skill character budget now scales with context window (2% of context), so users with larger context windows can see more skill descriptions
     without truncation
     • Fixed Thai/Lao spacing vowels (สระ า, ำ) not rendering correctly in the input field
     • VSCode: Fixed slash commands incorrectly being executed when pressing Enter with preceding text in the input field
     • VSCode: Added spinner when loading past conversations list

Re: Claude Opus 4.6

#63

Will Opus 4.6 via Claude Code be able to access the 1M context limit? The cost increase by going above 200k tokens is 2x input, 1.5x output, which is likely worth it especially for people with the $100/$200 plans.

The 1M context is not available via subscription - only via API usage

Re: Claude Opus 4.6

#64
post #25

Earlier quoted context omitted.

Ted Faro, is that you?!

A-tier reference. For the unaware, Ted Faro is the main antagonist of Horizon Zero Dawn, and there's a whole subreddit just for people to vent about how awful he is when they hit certain key reveals in the game: https://www.reddit.com/r/FuckTedFaro/

Average tech bro behavior tbh

Re: Claude Opus 4.6

#65
post #28

I found that "Agentic Search" is generally useless in most LLMs since sites with useful data tend to block AI models. The answer to "when is it cheaper to buy two singles rather than one return between Cambridge to London?" is available in sites such as BRFares, but no LLM can scrape it so it just makes up a generic useless answer.

Is it still getting blocked when you give it a browser?

Re: Claude Opus 4.6

#67
post #40

The bicycle frame is a bit wonky but the pelican itself is great: https://gist.github.com/simonw/a6806ce41b4c721e240a4548ecdbe...

What about the Pelo2 benchmark? (the gray bird that is not gray)

Re: Claude Opus 4.6

#68
post #17

> We build Claude with Claude. Our engineers write code with Claude Code every day well that explains quite a bit

Also explains why Claude Code is a React app outputting to a Terminal. (Seriously.)

There’s nothing wrong with that, except it lets ai skeptics feel superior

Re: Claude Opus 4.6

#69
post #13

Somehow regresses on SWE bench?

i'd interpret that as rounding error. that is unchanged swe-bench seems really hard once you are above 80%

it's not a great benchmark anymore... starting with it being python / django primarily... the industry should move to something more representative
Post reply on HN