I'm honestly not too aware of the true scene of which optimizers are used nowadays by which labs (though I think Moonshot has shown Muon's ability to scale to large models, and Muon is definitely more used now). That being said, obviously orthogonalization is pretty expensive! I looked it up and it's roughly 5-15% slower per step, but it converges in fewer steps fwiw. I'm guessing it's also less battle-tested than AdamW is, so AdamW still might be the safer option (and you're gonna have to use AdamW for your 1D params anyways).

I do think there are now open weight models that are on par with (or beating) Opus 4.5 by now (e.g. Kimi K3, GLM5.3). But yeah obviously the frontier closed source models seem to have pulled away once again, so open weight seems to be a few months behind right now (which might be too long to wait for a lot of people!).

Those two you mentioned completely demolish opus 4.5. It's not even close. I'd say they are between opus 4.8 and opus 5. And better in some tasks.

And if you run it yourself, you know it will be generally consistent with no random quality dips.

Opus 5 feels like a regression compared to 4.8

Olmo is one truly open source model. https://allenai.org/blog/olmo3

You can't ingest Common Crawl and claim to be an Open Model. Common Crawl is just a premade collection of random copyrighted unlicensed content.

I think it's just a harness that uses proprietary (or maybe IBM is hosting open weight?) models. They say on the site (in the FAQs) that you can't even choose which model!

So it seems to be a harness that auto-routes your queries with no way to change this.

You are correct. It routes to the model best suited for the ask.

The only positive versus others is that at an enterprise level you have more control of the costs with the coins.

except they tack on surcharges to all token calls to Claude and GPT

IBM has models trained for older mainframe usage (RPG II, RPG/400, COBOL, etc) so I imagine it would be very good for that. Bob looks like a branded fork of vs code or vscodium.

It's pretty incredible how robust neural networks are to even things like architectural changes mid-run. The team at Meta when training their own copy of GPT-3 (called OPT-3) even changed optimizers mid-run (from AdamW -> SGD -> AdamW, https://arxiv.org/pdf/2205.01068 section 2.5)!