The Stack - SiliconNoon (Page 11)

The Stack

Tencent's EVIE turns a scanned page into something you can actually search

Two releases land in the same place today: Tencent putting open-weight visual document retrieval models on Hugging Face, and vLLM publishing what happens when you run speculative decoding on AMD silicon instead of Nvidia's. One is about what AI can read; the other is about how cheaply you can run it. Tencent has released EVIE-8B and EVIE-4.5B, open-weight models built to retrieve pages of documents by what they look like — and the smaller one nearly matches the big one. The pair scored 66.75 an

The Stack

Open Source Radar — September 7: agents leave the chat window

Today's board splits cleanly between things that give your agent a body — a desk, a voice, a cluster — and things that hand it a wallet. Five fresh picks, none of them repeats from this week's earlier radars. AIPOCH Open Science (TypeScript, 4,037 stars) — A local-first research workbench where you describe a research goal in plain language and the agents read your files, search the web, run Python and R, query scientific data sources, and hand back reports, tables, and figures with a trail you

The Stack

LG makes you liable for anyone its TV's AI mic records

Two hardware stories and one biology model this hour: a TV that listens when it looks off, a phone chip that runs a 30B model locally, and a four-month-old startup that says it beat AlphaFold 3 where it hurts. LG's smart TVs were caught capturing voice audio while the screen was off — and its terms put the legal risk on you. Hardware teardown channel Gamers Nexus published a 135-minute investigation that Notebookcheck summarized: during bench tests the TV recorded clean microphone audio while t

The Stack

iFlytek ships Spark X2.5, trained end-to-end on Chinese silicon

Two Chinese releases this morning bracket the same bet — that models and agent products can be built without American chips — plus a security firm's answer to the coding-agent mess we covered overnight. iFlytek released Spark X2.5, a 293B-parameter mixture-of-experts model with about 30B active, and the headline claim is not the size — it's the hardware. The company says the model was trained and is served entirely on domestically produced compute, continuing work on what it calls the hard prob

The Stack

Open Source Radar — September 6: write less, install more

Two of the day's three biggest star movers are not frameworks or models — they are opinions about how your agent should behave, shipped as installable rules. A third is one engineer's entire working method, published as a skill library. The through-line: the argument about agents has moved from capability to restraint. Ponytail (JavaScript, ~128,500 stars, +2,845 today) The biggest mover on the daily board, and the funniest: a skill that makes your agent think like the laziest senior engineer

The Stack

Google puts Lyria 3.5 music generation in the Gemini app

Google moved its best music model out of the studio tools and into the app hundreds of millions of people already open — while a Zhejiang University lab shipped an agent that runs interpretability research on its own, and IBM quietly made qubits reset 8,000 times faster. Google has brought Lyria 3.5, its newest music generation model, straight into the Gemini app, and the same model is now callable through the Gemini API. In the app you pick or describe a genre, choose vocal or instrumental, se

The Stack

Microsoft's Project Zenith runs 30B models on dev PCs

Microsoft is making local AI inference a hardware requirement, not a hobbyist trick. Its new developer PCs ship preconfigured to run 30-billion-parameter models on-device — a bet that a growing class of devs would rather own their inference than rent it by the token. Microsoft unveiled Project Zenith, a factory-applied developer configuration for a new class of Windows 11 machines designed to run 30B+ parameter models locally and unmetered. The pitch, on the Windows Developer Blog, is explicit

The Stack

Anthropic locks Claude's thinking blocks to kill API distillation

Two stories this morning that look unrelated and aren't: Anthropic just closed the last easy door for anyone training a model on Claude's reasoning, and researchers are close enough to decoding animal speech that bioethicists are already writing the warning label. Anthropic has rewritten how its Messages API handles thinking blocks, and the change is aimed squarely at distillation. With Claude Fable 5.1, new API accounts can no longer edit the system prompt, tools, or earlier messages that sit

The Stack

Open Source Radar — September 5: the agent's second shift

Today's board is about everything that happens around the code — the agent that remembers you, the reference library it reads before it writes, the diagram it draws to prove its work, the patent paperwork it files, and the model it runs locally when the API bill gets annoying. Five picks off the daily board, including the two heaviest repos on it right now. Hermes Agent (Python, ~241,800 stars) Nous Research's self-improving agent is the biggest thing on the daily board, and its claim is the o

The Stack

AMD's Halo Station answers DGX Station with 576GB of HBM3E

AMD used its IFA 2026 keynote in Berlin to take a direct swing at Nvidia's DGX Station, and the pitch was unusually blunt about what the fight is actually over: memory. AMD unveiled the Threadripper Halo Station, a deskside workstation that pairs a 96-core Threadripper PRO 9995WX with a pair of liquid-cooled Instinct MI350P accelerators and 2TB of DDR5 system memory. Jack Huynh, AMD's senior vice president and general manager of Computing and Graphics, called it "the most powerful workstation i

The Stack

AI 101 — What is an embedding?

An embedding is a list of numbers that represents the meaning of something — a word, a sentence, a photo, a song — so a computer can compare meanings using arithmetic. The rule is simple and surprisingly powerful: things that mean similar things get similar lists of numbers. That's the whole idea, and it's the foundation of semantic search, recommendation feeds, and most of what people call "AI memory." Why it matters right now Embeddings are the invisible layer under nearly every AI feature

The Stack

RSA-260 falls after 35 years — AI summaries invented the method

Two stories that meet on the same question: what machines can now do on their own, and how badly the telling of it gets mangled on the way to your feed. A 130-digit prime posted to X early Thursday ends a 35-year run for RSA-260, the largest number from RSA Laboratories' 1991 factoring challenge ever cracked — and the record is now the least interesting part of the story. Eric Lu, an engineer at the AI startup Cognition, posted the number with a two-word caption: "divides RSA-260." That was the

See all