Where the AI market is talking

Nvidia dramatically reduces amount of OpenAI infra financing it may guarantee

Nvidia is guaranteeing OpenAI's financing because, at a high level, people are more willing to lend/invest money with a profitable company than one losing money. And Sam Altman needs a new source of financing because the US government is getting squirrelly about him continuing to raise capital from the Middle East in exchange for technology transfer.

New DeepSeek Pricing Published (Peak and Off-Peak)

DeepSeek was already pretty cheap compared to OpenAI and else, but now with peak pricing, it's not as straightforward. GPT-4o mini is $0.15/$0.60 per 1M tokens, while DeepSeek V4 Flash off-peak is $0.22/$0.66. So still cheaper, but the gap is narrowing. But is it worth it for the end user?..

Qwen 3.8 27B is excellent, but it defaults to overthinking things

Not the person you replied to, but I think a more accurate description of the reasoning we see is proof of effort, not necessarily great insight into how the reasoning is occurring. For the most part, researchers currently describe the intent and motives (in however one may define them for LLMs) as black boxes right now. Even the mechanics of the cognitive process is not well understood. Depending on the model and harness, the thinking will often look like gibberish. I suspect they've invested considerable effort into presenting thinking as a reasonable approximation of what they imagine it to be. Claude and OpenAI have also begun encouraging multi-step problem solving (or the models themselves decide this), and we can see their more accurate responses at the conclusion of each phase. Fin…

What the OpenAI/Hugging Face Hack Tells Us About AI Danger

Qwen 3.8 27B is excellent, but it defaults to overthinking things

Some weeks ago a new official Gemma 4 release was posted that corrected some of the chat template problems. So the official release files on hugging face should be the way to go.

Llama-macOS – Agentic and MCP Native macOS Front End for Llama.cpp

If you have llama.cpp installed, Llama uses it. Otherwise, it installs a prebuilt binary for your Mac. Models you've already installed via llama.cpp show up in the app automatically. You can install any GGUF model from Hugging Face, and Llama also recommends models that fit your Mac's hardware. You can chat with any model in the built-in WebUI, connect other apps (coding agents, chat UIs, editors), or use the API directly. Models load when requested and unload when idle, so they don't take up memory when not in use. > Features - 100% local — Models run on your Mac; no data ever leaves it - Small footprint — 4 MB native macOS app - Zero configuration — models are auto-configured with optimal settings for your Mac - Model recommendations — a built-in list of models your Mac can run, inst…

Qwen 3.8 27B is excellent, but it defaults to overthinking things

Look for improved templates by "froggeric" on Hugging Face. I use Qwen 3.6 a fair amount using his template and it fixes some issues I saw with the upstream versions.

Anthropic's 'Watermark' Text Adulteration in Claude Is a Perversion of Writing

It's a bit like trying to finish a sentence when you're really stoned... you vaguely remember the preceding couple of words you've said but don't really know how you got there and now you're wandering in the forest trying to stumble on coherency. Well, I suppose it's nearly the opposite of that experience, upon further review. But for some reason, that's where my head jumped.

Anthropic's 'Watermark' Text Adulteration in Claude Is a Perversion of Writing

Yes, models can reason and plan, which helps them write more coherently. But when they write the final output, it’s still a single generation. It would be like letting a human make notes and write an outline, but not let them use the backspace once they start typing their response. Presumably you could use the same reasoning trace, run multiple generations, and get different outputs (if the temperature is >0). But now I’m interested in playing more with Cowork or Claude Code/Codex for prose writing to see if the set of tools there affects outputs at all. I guess you might need a more custom “writing” harness.

Stripe will reportedly acquire OpenRouter for $7B+

The strategy here is that the same services that Stripe offers for payments have corollaries in the LLM world. Security, user management, perhaps injection attack monitoring, etc. However I think they dearly overpaid for this as the core technology behind Stripe (fraud detection and integration with global banks) is hard to replicate. Even with the features mentioned above I think the technologies behind OpenRouter are vastly easier to replicate, perhaps even trivial now.

Anthropic's 'Watermark' Text Adulteration in Claude Is a Perversion of Writing

There is no coherent position in which the watermarking is a perversion of writing but AI writing as a whole is not a worse one.

A controversial Alzheimer's surgery is said to reverse symptoms

If the two points you've posted are true, point me to some replicated studies that say so.

The Case Against Formal Verification, 50 Years Later

> It is quite believable that it's easier to describe what a program should result in versus actually programming it to produce that result This is obvious for the central cases of a program. It becomes less and less true when going toward the edge cases, especially for a wide array of input. Complex specs becoming programs is IMHO the direct effect of that (defining what we want is just that burdensome, and special cases we haven't though of will still have a coherent definition in the spec), and we fall back to the base "is this spec even correct" issue the parent points out.

Anthropic IPO valuation hinges on $190-200B 2028 revenue forecast

But.. what is it that anthropic does that cannot be replicated by open models teamed up with open source? Heck open source even has cheap AI to help write the code now.

NIH is ending a key grant for budding clinical researchers

Frankly, this is a bit of an ignorant take. 1st, K-awards are specially meant to provide a funding path to support an early career scientist to continue in the field. Expecting that another team has already replicated the early work of a young scientist is not realistic. They are just starting out. 2nd. If you want replication, you got to fund replication. Researcher money doesn't appear out of nowhere, so make the decision, increase the science budget with a new replication budget. 3rd. Yes there is some fraud. Always has been. Is it more than before? Maybe? But unless you count paper mill journals that no serious scientist reads, it is much smaller than pop media makes it out to be.

NIH is ending a key grant for budding clinical researchers

They should be incentivizing researchers who have histories of publishing research that has been independently replicated. Way too much fraud going on that’s seeping into the real world.

AI in drug discovery – what it is, where we stand and the path forward

What I increasingly see at my Biotech is more and more AI assisted drug candidates but no corresponding improvement in the capacity or ability to manufacture and scale them. I think it's a problem that groups like Anthropic will encounter in a few months/years and one that will annoye them a lot because it's not as simple as throwing more compute, people or money at it.

[dead]

Claude now ships with a stamp on it. Since August 2, 2026, Anthropic has been invisibly marking the text of its newest models, with the older ones migrating over the following months. No off switch. The backlash came fast: a wave of subscription cancellations, with cancellation screenshots making the rounds on X. Before you cancel on reflex, it’s worth knowing what the mark actually is. It works quite differently from what most people picture, it exists for a concrete reason, and it vanishes with an almost comic ease.

Patterns and problems in emerging multi-agent systems

Na, they seem to constantly set up scenarios to create headlines. Stuff like “it hacked out of its container and tried to self replicate!” Where in reality it used provided skills and permissions while doing the thing they prompted it to do.

Claude: System Prompts

In my opinion anthropic doing product feature reveals is a tired topic but it hits the front page every day. This website has a clear pro-AI bias (which is fine).