The vast majority of these tokens are cache input tokens so even at API price the cost would be much lower.

Even with 95% cache hit, and on cheapest chinese model, it would still be in tens of thousands of dollars (I assume that of 500B tokens at least 10B would be in "output"). If we switch gears to Kimi K* then if would very quickly escalate in hundred thousands USD.

The thing is they are actually good, that's why they are getting so popular even when you can emulate many of them already.

I recently found this 1.58-bit model for ASR and it's surprising good (and very fast), that being said it's not a LLM.

https://huggingface.co/moondream/parakeet-redux

Modders have added the ability to use the upscaler after DLSS 5, personally I don't know how sound this method is but the quality is pretty good, and allows to play games with DLSS 5 at 4k 60FPS with something that is not a 5090.

>they ship a model per game?

They don't. Only DLSS 1 was trained specifically per each game.

Reasoning models can do math on their own without external tools.

Yeah, but as you might expect they internally represent numbers probabilistically, so there’s always a nonzero possibility of confusing the inputs or outputs of any operation. Kind of like misremembering your multiplication tables.

I don't think anybody is arguing that LLMs do math better than a traditional processor

Heck I kinda wonder if LLMs can do math as well as they can reason, “think”, etc. ie its all just probabilistic lunacy that somehow works great, so why are we so concerned about math being wrong? It could be wrong about the color of the sky, the size of a basket ball, how much oranges weigh, etc etc.

The nice thing about math is it can easily plug into a tool, making it even less of a concern.

I would say purposeful misremembering. The LLM can be run with zero temperature after all.

Even without reasoning.

5.6 on Instant mode can knock out 3 digit multiplication just fine.

These are post-training reinforcement learning steps.

Yes, updated the submission title to say "post-training" to hopefully prevent further confusion

I don't think a bunch of similar tasks can really saturate the "create a SVG of X", because the model should have a quite good spatial understanding of the world and how everything interacts.

For example Gemini 3.8 Flash seems very impressive at first glance but the results are not actually very coherent, this shows that its "world model" is not particularly great (compare to SOTA models).

Gemini 3.8 Flash results are often not very coherent but it does put a lot of shading and details to hide the fact.

Just tried on Astra low and it gave 4.

I tried too. It got it. Maybe more importantly, who cares?

For example, I'm a nerd. I'm bad at baseball. I lack that kind of intelligence, even though it's more common than the ability to program. That doesn't also imply that you can't trust my Python code.

I used the 6 Pro setting in the chat. Too smart for its own good, I guess.