amazing to see Claude Code top models still way above all other models for C++ & Java, while GPT 5.5 is higher in Python & JS and others. Shows the skew in the training data sets, and maybe the go-to-market focus - with Anthropic focusing on enterprise customers much more than OpenAI?

Matches with my experience with Opus for C++.

C# results are empty - @gertlabs - any ETA for those?

C# testing is a new feature added a few days ago from HN comment suggestions, samples will continue growing. Most C# data is currently for non-agentic workloads: https://gertlabs.com/?mode=oneshot_coding

Does this include repos content in BitBucket?

For those in the USA - if you never tried brunost (brown cheese) - look at specialty cheeses of a larger grocery chain (Whole Foods, Safeway, …) - it’s called “Ski Queen” here, and is sold as a perfect cube in red/brown plastic pack.

It’s very delicious.

I was ecstatic when I found it quite a few years ago in a regular store. A Norwegian friend of mine used to send me a brick of this cheese once a year for Christmas, when I was a student, and I treasured it as one of the most valuable possessions :)

Also, fun fact - the reason this cheese tastes sweet is due to caramelization - the milk gets boiled for a long time (hours) to get the brown color and sweetness. So it’s completely natural, zero added sugar ;)

He may just want to buy them, to accelerate things, once SpaceX IPOs

Fwiw, this sounds like a healthy discourse - you don’t have to agree on everything, every approach has its merits, code that ends up shipping and supporting production wins the argument in some sense…

This is not special to Meta in any way, I observed it in any team which has more than 1 strong senior engineer.

No, calling out your ex colleague in public years after is not a "healthy discourse" ...

There's nothing healthy about holding on to a work grudge from 10 years ago and then dragging it out in public. That's toxic AF.

Tangentially, on this CD policy - it leads to really high p99s for a long tail of rare requests which don’t get reliable prewarming due to these frequent HHVM restarts…

very cool idea. But, time savings are not true for every tool call, and it's not clear to me yet whether this is batch-able; also, intuitively, for most of the models that run on GPU, you'd still want to offload tool exec part to CPU since it's much cheaper...

> I'm not fully on board with his world model strategy as the path forward

can you please elaborate on your strategy as the path forward?

is Google using LLM-guided fuzzers that can inspect the code first?