sharmajai

Yes, that's what I meant, thanks for clarifying!

If you want to draw from history there are multiple successful examples of this in science and engineering at least: Bletchley Park, the Manhattan Project, Moon Landings.

Capital allocation to advance mathematics and sciences is undoubtedly a good thing. But this does beget a few questions:

1. How much did it cost?

2. Would an equivalent allocation to incentivize human effort have borne the same results?

3. Post IPO, will Open AI continue allocating similar levels of capital to further these causes?

#2 is particularly important to answer, given research grants have been getting cut elsewhere: see U.C.L.A's research grant cuts from last year and Terrence Tao's attempt to raise funds (about $60k) by collaborating with Open AI earlier this year [1].

[1] https://www.nytimes.com/2026/10/16/us/math-ai-terence-tao.ht...

A cloud model can also get neutered anytime and/or start using more tokens while charging you the same or more per token.

The answer my friend, is blowing in the wind (https://www.youtube.com/watch?v=MMFj8uDubsE&list=RDMMFj8uDub...).

I'll let the data do the talking:

Cursed companies:

- https://permtrack.app/employers/microsoft-corporation

- https://permtrack.app/employers/tata-consultancy-services-li...

Blessed companies:

- https://permtrack.app/employers/oracle-america-inc

- https://permtrack.app/employers/tesla-inc

I see, maybe your use case is indeed pathological for Qwen 3.8 27b. Another thing to try if you are using llama.cpp is "reasoning budget". That makes the thinking stop after the budget has been reached and inserts a custom message you can choose, so something like "you have thought for too long, now continue with the execution ..."

For what it's worth, if you haven't already, you can also let it run overnight (if you have compaction enabled) to see if it ever gets out of that hole. The reason `xhigh` is the default is because 3.8 is trying to optimize for long horizon tasks where monitoring its every single thinking misstep might not be a good use of our time.

Without knowing anything about your benchmark, it might be your harness at fault here. I say this because with longer thinking you run the risk of filling up your context faster and you need a good context compaction strategy in your harness to mitigate that.

I have heard good things about Pi which supports auto-compaction, but I can't personally vouch for it since I use my own.

This confirms a theory I have to explain the minimal loss in quality when using lower quants (I use IQ3_XXS with an 8-bit KV cache) and the XHIGH (default) thinking level.

It's well-known that while quantization affects the sampling probability distribution (given the same context, which next token is the most probable), Qwen 3.8 27b seems to offset that by just thinking more and as a result eventually finishing the task (benchmark or otherwise).

So as long as the thinking (albeit longer) is sound, this leads to the same success rate (as shown in the article) but potentially at the cost of more tokens and hence more time.

I think it'll be further useful to chart each quantization's used tokens as well, in addition to the success rate.

Thanks for doing and sharing the research!

Really feels like AGIPO is here.

Thanks for the movie suggestion, looks very interesting, will check it out. Also, the Pixar movie Soul has a similar message.

You'd be a great CEO, preaching people to tie their worth to their output in life.

I am getting 14 t/s on my 16 GB card at full context with the UD-Q3_K_XL quant. Model link: https://huggingface.co/unsloth/Qwen3.8-27B-GGUF.

Who said investing is _only_ for "capital and capital appreciation"? It can also be for social good.

They have a very opaque and subjective permanent residency program. So while they get all the benefits out of you as an immigrant, they may provide none in return.

Product companies with deprioritized R&D wings are the first ones to die.

I get the urge to be cynical all the time, but this isn't that time. "Once you grow", they have already grown and competing with the SoTA models and still giving it all back to the community.

I just wish this smear campaign against them stops sometime soon.

I am 100% sure that company is the F in FAANG.

Dude, that's cheating!

But but look at all the Turing Award and Putnam Prize winners he was worked with.

Maybe not everything should be about business.

What're the issues with PayPal that you're aware of?

That's being tried [1]. Although, it's not going well [2].

[1] https://www.visalaw.com/onboarding-mass-litigation-clients/

[2] https://twitter.com/LilySAxelrod/status/1437846756394520582

While I agree that Peter's suggestion is the safe thing to do, based on the following tweet from the official account, if you were in the country legally on 06/24, the proclamation doesn't seem to apply.

https://mobile.twitter.com/TravelGov/status/1285331446232743...

Of course it does, because it can be combined with other characters. This is the semantic meaning: https://en.wikipedia.org/wiki/Acute_accent

I don't see which of those 4 definitions supports the grapheme cluster interpretation.

<nitpick>All Unicode characters map, one-to-one, to their code points. A code point being a numeric identifier. It's a grapheme that combines multiple characters to form a unit of writing.</nitpick>

That's an amazing price given it's retailing for more than twice that. One option to expand storage is have an always-inserted, low-profile USB 3 flash drive like SanDisk Ultra Fit, providing read speeds upto 150 MB/s.