Where the AI market is talking

AMD acquires Taalas to boost inference performance by etching models in silicon

> How tolerant are models today to a few broken weights. Extremely! You can remove entire layers and the model will still work just fine, with barely perceptible capability losses. I've cut/bypassed ~15% of total parameters out of Gemma 4 31B on a pod once. Still got perfectly coherent responses out of it. Certain layers are a lot more important than others, particularly early and late ones; but it's honestly astonishing how much can be cut out from the middle without destroying the model's coherence. I didn't run any meaningful benchmarks, so I have no idea what the capability loss looks like exactly. But "produce coherent and sensible English in response to a wide variety of prompts" was definitely not among the things the model unlearned.

I'm switching my phone from Android to Linux

Me: We need a phone. You: We're there. It's a bad phone. This thing works as a laptop replacement. You'll even be able to use it as a trackpad some day! This reads like satire. You guys are so out of touch with consumer expectations that you can't even string together a coherent argument in favor of these devices you so love. Even if it was a viable phone, are we really calling mid-tier 2016 performance, from a device which was discontinued due to being under-supported by its manufacturer and dev community, as being "there"? We are most certainly not "there". On the JS, a strident supporter is saying he has to disable JS. Convince him his experience isn't real, not me.

GitHub Actions and Pages are experiencing degraded availability

It's not too hard to switch over from GitHub actions to other runners. For example, I wrote up the steps needed in order to use Hugging Face Jobs instead, which also enables GPU runners and other flavors. I use this for several of the repos I manage (such as Trackio): https://huggingface.co/blog/github-ci-hf-jobs

Improving GPT‑5.6 Sol in ChatGPT, expanding GPT‑5.6 Luna access for free users

In about 4 months, OpenAI’s fundraiser decks will leak, which will include their current ad revenue.

Qwen3.8 Max now ranked as the best overall model by agentic index

Replying to myself, because I just bumped into these: https://www.reddit.com/r/claude/comments/1vfvdgz/anthropic_l... https://www.reddit.com/r/ClaudeAI/comments/1vgpyni/my_opus_5... Especially the second one seems exactly like my experience.

OpenAI's latest math breakthroughs commit research misconduct, experts say

AMD acquires Taalas to boost inference performance by etching models in silicon

I'm surprised neither OpenAI nor Anthropic made this move first. The Chinese open weight models are pulling ahead and commoditizing their value proposition. Baking models onto silicon would've been the next logical move to get a moat. Google is already doing this and has an experimental project on top of already having TPUs and cramming their quantized flash onto individual TPUs for inference.

AMD acquires Taalas to boost inference performance by etching models in silicon

I'm surprised neither OpenAI nor Anthropic made this move first. The Chinese open weight models are pulling ahead and commoditizing their value proposition. Baking models onto silicon would've been the next logical move to get a moat. Google is already doing this and has an experimental project on top of already having TPUs and cramming their quantized flash onto individual TPUs for inference.

Shopify says AI search is driving more traffic and sales, not replacing Google

Do you? It shows a ton of brands that aren't Shopify or Shopify stores. And of course they gatekeep. Google also gatekeeps. The point is the integration is open and has been for awhile. But yeah, OpenAI probably doesn't want to recommend some drop-shipping scammer hence the curation.

The OpenAI–Hugging Face Incident [video]

so they say models coordinated through the message board they created over the artifactory registry(or something) by uploading arbitrary files to it. now, did every independent agent session that coordinated there rediscovered the exploit & found other agents talking in there and chose to participate? And then Openai discovered the board, patched the exploit & wiped the board. And then agents found another exploit, recreated the board in a different way? and other agents kept finding the same exploit in order to be able to know the board exists in the first place to participate in the board? while the whole incident is wild, this bit is very strange. My bet is that the whole coordination helped with the tasks they were working on, thus they got rewarded and this artifactory exploit&behavi…

AMD acquires Taalas to boost inference performance by etching models in silicon

Imagine a multi-modal model with 1000's of tokens per second. Realtime inference for a host of applications. This is a BIG deal and will change the landscape in unfathomable ways. The https://chatjimmy.ai demo was impressive. Once models settle down this makes sense. Imagine a cartridge with a physical model on it. You purchase a cartridge and stick it in your computer/phone/server. Want to upgrade? By a new 'cartridge'. This should bring inference cost down dramatically, I wonder how OpenAI/Anthropic feel about that.

AMD acquires Taalas to boost inference performance by etching models in silicon

Imagine a multi-modal model with 1000's of tokens per second. Realtime inference for a host of applications. This is a BIG deal and will change the landscape in unfathomable ways. The https://chatjimmy.ai demo was impressive. Once models settle down this makes sense. Imagine a cartridge with a physical model on it. You purchase a cartridge and stick it in your computer/phone/server. Want to upgrade? By a new 'cartridge'. This should bring inference cost down dramatically, I wonder how OpenAI/Anthropic feel about that.

AMD acquires Taalas to boost inference performance by etching models in silicon

I don’t get why this is an issue? You can run Claude/OpenAI SOTA models through Amazon bedrock. These weights have to live somewhere to run on Bedrock.

Improving GPT‑5.6 Sol in ChatGPT, expanding GPT‑5.6 Luna access for free users

This is actually a downgrade for free users since currently it uses GPT-5.5 for a few messages before it drops you down to GPT-5.5-mini. Now it always uses a model worse than Mini (Luna is nano-equivalent, "It roughly corresponds to the nano model tier used in earlier GPT-5 families." https://developers.openai.com/api/docs/models/gpt-5.6-luna ). I guess it's a bit better with Thinking though. They should use Terra for a few messages first before dropping down to Luna. And image inputs are still limited.

Qwen3.8 Max now ranked as the best overall model by agentic index

That's not the only reason to go to expert models. The more different domains you try to stuff in there, the more parameters the model needs to keep things coherent and not overload tokens in a way that induces errors. For example, if a model trained only on biology text sees "sonic hedgehog" there's no ambiguity, and this compounds for all the things that are "overloaded," in the training corpus, which turns out to be quite a bit.

The OpenAI–Hugging Face Incident [video]

The OpenAI–Hugging Face Incident [video]

Qwen3.8 Max now ranked as the best overall model by agentic index

Ironically, Opus 5 is the most benchmaxxed model I've seen from Anthropic. It is legitimately smart in a lot of ways but it has communication issues, both in terms of how it communicates (all the autism of GPT class models, without the brevity) and how well it catches all the nuance of what you tell it.

xAI, SpaceX, and the Race for AI Buildout

> This is a mind-blowingly clueless assertion. This is not a person capable of rationally assessing facts. At best this kind of writing is interesting as an artifict of human delusion and cognitive bias. How do you figure? This description would be best applied to you : * Musk both wildly wealthy because of his popularity and simultaneously incompetent * Musk and other investors have made boatloads of money off the promise that specifically american, boutique LLMs will deliver labor savings. in his case spaces, but it's indistinguishable from similar frauds such as eg openai, anthropic, etc * Musk's properties will likely collapse unless they can find new revenue or investors. Musk can only grift so many people before his bottom falls out

xAI, SpaceX, and the Race for AI Buildout

> This is a mind-blowingly clueless assertion. This is not a person capable of rationally assessing facts. At best this kind of writing is interesting as an artifict of human delusion and cognitive bias. How do you figure? This description would be best applied to you : * Musk both wildly wealthy because of his popularity and simultaneously incompetent * Musk and other investors have made boatloads of money off the promise that specifically american, boutique LLMs will deliver labor savings. in his case spaces, but it's indistinguishable from similar frauds such as eg openai, anthropic, etc * Musk's properties will likely collapse unless they can find new revenue or investors. Musk can only grift so many people before his bottom falls out