Any other details?
Fallback was usually Opus 4.8
Isn't that a worry then that the same bench has so much difference in what triggered fallback for one model and what did not in another?
Do we have a reliable way to count tokens for Jev yet btw?
Slop projects to solve slop
Does the price really matter when you are on subscription? Are we getting more usage or are we getting same usage and the cost for openai is lower?
He also blogged why he thinks it's still useful
Which is actually good, as it includes GST.
It includes GST right? US prices are before sales tax.
Luna, Terra, Sol, Astra. Though Sun is also a star, should have called it Galaxy or something.
Weird phrasing "just a toy" for one and "hacker toy" for the other. Beni is actually something you can use more than a toy.
Is it Canadian dollar for US dollar? /s
If you re getting dumber, you would feel like it's all good right? Why would you feel Opus 4.6 is better than Opus 5
Python, but good.
Claude Code? It stores a memory.md file.
Opus 4.8 was already shown to be cheaper than Sonnet 5 when Sonnet 5 was released (by Anthropic)
Which numbers are you seeing? It does show that it's better than Fable 5 in most things related to coding?
Yeah and Bun and Zig are not competing in anyway as well. Zig blog post has been updated as well recently btw.
Because they want a Haiku, Sonnet, Opus equivalent I guess?
Why the name Golem?
Don't judge a book by its cover. In this case, a library by its website.
I was talking about enterprise agents and then realized the question is more about coding agents.
Which is your own harness and your own evals for your tasks I guess
Opus 4.8 high doing better and cheaper than Sonnet 5 xhigh
All a facade no? It's just about regulatory capture.
This comment reeks of someone who has not used Cursor, at least in 2026.
Fully on Cursor at work and I love it over CC, OpenCode and Pi that I use for personal work.
Composer is very good, and these days after heavily using CC, Pi and Open code I am back with Cursor. "No moat or IP" is underrating it a lot.
But many do already own SSD and RAM that they can reuse.
This is not 2015.