Qu

ML research | 30.05.25 🫶🏻

[Qu @nigo_ai]

The function of man is to live, not to exist.
I shall not waste my days trying to prolong them.
I shall use my time.

[Hedgeye @Hedgeye]

Global bond yields rise to highest level since 2008

[Qu @nigo_ai]

a new world is being born

[NIK @ns123abc]

🚨BREAKING: China has begun mass production of DUV lithography machines for the first time

ASML: nobody can touch our monopoly

China: targeting 5 DUV machines this year and 20 next year

Chip equipment stocks:

ASML -8.3%
ASM International -7.3%
BE Semiconductor -9.9%
Applied Materials -6.7%
Lam Research -7.9%

It’s over

[Pedro Domingos @pmddomingos]

Excellent idea. Universities are just infrastructure providers, and every day we need it less.

[Andrew Curran @AndrewCurran_]

In an new report expected to be released later today the Trump administration will argue that directing federal funding towards individual researchers, and research that uses AI, will accelerate scientific progress faster than funding collages and universites.

[Barchart @Barchart]

BREAKING 🚨: Japan

Japan's 2-Year Bond Yield soars to 1.5%, the highest level in more than 30 years 🤯 👀

[Kalshi @Kalshi]

JUST IN: Fidelity buys $23 million worth of Bitcoin

[International Cyber Digest @IntCyberDigest]

‼️ UPDATE: LG Electronics says it will suspend smart TV apps that secretly turn hundreds of millions of TVs into residential proxy nodes, after researchers found the proxy SDKs in more than 42% of apps in LG's webOS store and over a quarter of Samsung's Tizen apps, routing unknown third parties' internet traffic through users' home connections.

Residential proxy firms pay developers to bundle these SDKs, then rent the resulting home IP addresses to customers who often use them for large-scale scraping aka stealing your shit (also for AI training).

Proxy nodes are being embedded at scale in hardware people don't treat as computers and can't easily inspect, and LG's cleanup does nothing for Samsung's Tizen store, where the same problem is present.

[clem 🤗 @ClementDelangue]

Open-source is not the cause of the cybersecurity crisis, it's the solution! https://t.co/YpykmLGv7B

[Polymarket @Polymarket]

BREAKING: Google’s free cash flow turns negative for the first time in at least a decade as AI spending surges.

[Ahmad @TheAhmadOsman]

Fable 5 wasn't even available for majority of time Kimi K3 was being trained, any accusations of distillations CANNOT be true

[Polymarket @Polymarket]

JUST IN: GPT-5.6 Pro disproves the 30 y/o Dinitz–Garg–Goemans conjecture, a long-standing problem in mathematics — after being prompted to "do a breakthrough."

[Fireworks @FireworksAI_HQ]

We ran Kimi K3 against Fable on ~1,000 agentic tasks, expecting a catch-up story. We got a specialization story instead.

@kimi_moonshot's K3 outperformed on security, crypto, and long terminal loops. Fable beat on multi-lang + web/data viz. Per-task routing hits 93% accuracy, above BOTH models, at up to 50x lower cost than Fable on long loops.

The part nobody's pricing in yet: the router sends 72-96% of traffic to K3. The frontier model becomes the fallback rather than the default.

Kimi K3, coming to Fireworks July 27.

[Jun Song @jun_song]

moonshot team built a time machine and distilled from future Fable-6

[Global Markets Investor @GlobalMktObserv]

‼️Chinese AI models have officially overtaken the US globally:

China's share of AI token usage by US firms on OpenRouter jumped to ~58%, surpassing the US for the first time since this data began.

This share has more TRIPLED over the last few months.

Among US firms specifically, Chinese models now account for ~38% to 40% of tokens used, with DeepSeek remaining the single most popular choice, well ahead of Z Ai model, Qwen, MiniMax, and Kimi.

Meanwhile, Chinese startup Moonshot released its new Kimi K3 model last week, a release investors say rivals top US systems and one that triggered a fresh selloff in AI and semiconductor stocks, echoing last year's "DeepSeek moment."

Furthermore, Chinese officials are reportedly weighing restrictions on foreign access to their most capable models, facing the same security dilemma already playing out in Washington.

The AI race is no longer just about who builds the most powerful models, but who can make them cheaper, faster, and more widely adopted.

[Qwen @Alibaba_Qwen]

During Preview, Qwen3.8 is getting better by the day. Latest version is live now, with broad gains and a big step up on web frontend.

Thank you all — the response to Qwen3.8-Max-Preview blew us away. 🫶🫶

Qwen3.8 is still evolving daily. Come test it, and tell us what breaks.

We're looking forward to a more capable, official version — and to open-weight it for everyone.🚀🚀

[Kai @hqmank]

Kimi K3 vs Fable 5.

Same prompt and reference. 3D globe dashboard. Kimi left, Fable right.

To me, Kimi's UI fidelity is a touch better.
The globe looks better too, with richer textures and smoother motion.

This task: ~$3 with Kimi, ~$15 with Fable.
I'd pick Kimi k3 for frontend work.

[Zain @zainhas]

Kimi K3 on legal tasks is ~2x fable perf 🤯

The benchmark is from Harvey and involves hard autonomous legal work

Kimi K3 at 26.7%, vs Claude Fable 5 at 14.2%.

[Sakana AI @SakanaAILabs]

Introducing Fugu-Cyber: an update to our Fugu orchestration model.

It achieves state-of-the-art performance on real-world security benchmarks, matching cyber-focused frontier models like GPT-5.5-Cyber and Mythos Preview.

https://t.co/5Nh1eBPhHg 🐡

[Composio @composio]

We tested Kimi K3 against Claude Fable 5 on 14 agentic office tasks: syncing support tickets into spreadsheets, auditing GitHub repo access, deduping CRM contacts, and more.

Result: a perfect tie. Each model scored 9/14. They passed the same 9 tasks and failed the same 5.

An open-weight model just matched a frontier flagship model, case for case.

The differences showed up elsewhere:

Speed: Fable finished tasks ~2.5x faster.

Example: A calendar check took Fable 54 seconds, Kimi 200. A CRM cleanup took Fable 64 seconds, Kimi 274.

Cost: Kimi used ~40% fewer tokens across the full run. Fable burned 3.1M tokens on one task where Kimi used 1.7M.

[Hugging Face @huggingface]

Join us this Tuesday to learn more on Local AI, from software to hardware 🤗

We'll be joined by @TheAhmadOsman & @MikeBradleyAI covering hardware setups & local inference with live demo, and @alexocheema & @0xSero on picking model for your hardware, model compression and REAPs 🔥

Set your reminders to not miss out! 🔔