#strixhalo — Public Fediverse posts

ich scriptkidde mit #ornith-1.5:35b in #hermes in meinem linux system rum - das läuft ja lokal ganz fix auf meinem #strixhalo #localai

was mir direkt aufgefallen ist: der hat schon bei pacman systemupdate raten müssen... er hat es dann ja hinbekommen

aufgeben? niemals =)

denke hier ist die lösung: auch für banale dinge #skills erstellen lassen

#glm52 #glm53 hätte das direkt aus dem gedächnis hinbekommen - immerhin weiß #ornith den einstiegspunkt pacman =)

mal schauen wo das endet

GPD Win Max 3 mini gaming laptop with AMD Strix Halo to sell for $1750 and up at launch

The GPD Win Max 3 is a mini laptop with a 9.06 inch display, an AMD Strix Halo processor with discrete-class graphics, and a design that makes it clear that this is a little PC made for both work and play. It has a keyboard large enough for touch typing, but the touchpad is above the keyboard rather than below it, and it’s squeezed between a set of game controllers.

First unveiled earlier […]

ich benchmarke grade ein paar #localai llms auf meinem #strixhalo mit 200 GSM8K + 3×100 BFCL Samples

dabei ist mir aufgefallen das thinking moodelle wie qwen 3.6 0% ergebnis liefert - wegen zu langen thinking blocks

da in #hermes_agent #harness das thinking drin ist - ist das doppelt... und hat deshalb bei mir bei qwen3.6 immer wieder geloopt und war unbrauchbar

offensichtlicher unterschied: qwen3.8 hat thinking nicht aktiv! also ist das der gamechanger???

hier einschätzung #glm52
Thinking-Modelle in Agent-Loops: Das Problem
Die Recherche bestätigt unsere Benchmark-Ergebnisse eins zu eins:

1. Token-Budget wird im Agent-Loop multipliziert, nicht addiert
- Ein einzelner Query kostet mit Standard-Modell ~7 Tokens, mit Thinking-Modell ~255-603 Tokens
- In einem Agent-Loop mit 12 Iterationen zahlst du nicht 10x — du zahlst 10x × 12, und das wird bei jeder Iteration durch die wachsende History weiter amplifiziert
- Eine 10-Turn-Loop sendet ~50x mehr Tokens als ein einzelner linearer Call (Falconer Guides)

2. Thinking-Blöcke fressen genau das, was der Agent am dringendsten braucht: Context
- Stanford Digital Economy Lab: Agent-Tasks brauchen ~1000x mehr Tokens als Chat, mit 30x Variance
- Thinking-Output wird zur History hinzugefügt → bei jedem Turn ist mehr Context belegt → das eigentliche Task-Wissen wird verdrängt → schlechtere Ergebnisse
- Genau das haben wir gesehen: qwen3.6:35b mit Thinking = 1.5% GSM8K (fast alles leer), ohne Thinking = 100% auf 5 Samples

3. Reasoning-Loops sind ein echtes, beobachtetes Phänomen
- Ein dokumentierter Fall: ein Agent machte 847 Reasoning-Steps und lieferte nie ein Ergebnis (dev.to)
- "Overthinking leads to poor decisions" — jede zusätzliche Reasoning-Step macht es schlimmer, nicht besser
- Agent ruft dieselbe API 14x hintereinander, "nur noch ein Schritt mehr" wird das Ergebnis schon perfekt machen (ShopClawMart)

4. Reasoning bei reinen Action-Knoten ist reine Verschwendung

"Once a plan is formed and the current step is 'call the search API with this query' or 'write this string to that file,' reasoning tokens add nothing. The action is deterministic. A fast model will generate the correct function call as reliably as a reasoning model at a fraction of the cost. This is the most common budget leak in agent systems."

Was das für Hermes bedeutet
In einem Agent-System wie Hermes ist die Situation noch verschärft:
- Jeder Tool-Call erzeugt Output, der zur History kommt
- Der Context füllt sich ohnehin schon schnell (System-Prompt, Skills, Tool-Outputs...)
- Kommt jetzt noch ein 2000-Token Thinking-Block pro Turn dazu, ist der Context nach 5-6 Turns voll — und dann beginnt der Context-Overflow: ältere Task-Infos fallen raus, der Agent verliert den Faden, fängt an zu loopen
- Das ist auch warum du ja gesagt hast "Context fills up fast" in Hermes — Thinking würde das dramatisch verschlimmern

1/x

Statement zu KI und mein teilrückzug aus dem Fediverse

https://friendica.tf-translate.net/display/cafe12d9-116a-8634-5024-1d3980317993

Come Scegliere il Miglior Mini PC per l'AI Locale nel 2026: Strix Halo vs DGX Spark vs Mac

Un mini PC grande quanto un libro tascabile è oggi in grado di eseguire localmente un modello da 200 miliardi di parametri. Ma la scelta non dipende solo dal prezzo. La capacità della memoria determina quali modelli possono essere caricati, la larghezza di banda influisce sulla velocità di esecuzione e lo stack software — CUDA, ROCm o Metal — stabilisce se gli strumenti che utilizi funzioneranno davvero. Ecco un confronto tra le quattro principali opzioni disponibili nel 2026, con prezzi e benchmark aggiornati.

buysellram.com/blog/how-to-cho…

#AIlocale #LLM #MiniPC #HardwareAI #StrixHalo #DGXSpark #AppleSilicon #EdgeAI #InfrastrutturaAI #Ollama #RyzenAI #AIPC #AMD #NVIDIA #Apple

GPD Win Max 3 is a mini laptop for work and play with AMD Strix Halo and a removable battery

The GPD Win Max 3 is a mini-laptop that packs a lot of horsepower into a compact design. With a 9.06 inch display and a nearly full-sized keyboard and Precision touchpad, it’s a little laptop you can use for productivity on the go. But remove a couple of magnetic panels above the keyboard and you’ll find a set of joysticks for gaming.

And with an AMD Strix Halo processor featuring […]

A mini PC the size of a paperback can now run a 200B-parameter model locally. But choosing one isn't about the lowest price tag. Capacity sets what fits, bandwidth sets how fast it runs, and the software stack — CUDA, ROCm, or Metal — decides whether your tools work at all. Here's how the four real options compare in 2026, with current prices and benchmarks.
https://www.buysellram.com/blog/how-to-choose-the-best-mini-pc-for-local-ai-in-2026/
#LocalAI #LLM #MiniPC #AIhardware #StrixHalo #DGXSpark #AppleSilicon #EdgeAI #RyzenAI #AIPC #AMD #NVIDIA #Apple

Choosing A mini PC isn't about the lowest price tag. Capacity sets what fits, bandwidth sets how fast it runs, and the software stack — CUDA, ROCm, or Metal — decides whether your tools work at all. Here's how the four real options compare in 2026, with current prices and benchmarks.
https://www.buysellram.com/blog/how-to-choose-the-best-mini-pc-for-local-ai-in-2026/
#LocalAI #LLM #MiniPC #AIhardware #StrixHalo #DGXSpark #AppleSilicon #EdgeAI #AIinfrastructure #Ollama #RyzenAI #AIPC #AMD #NVIDIA

ONEXPLAYER X2 Mini Pro handheld gaming PC with Ryzen AI Max+ 388 hits Indiegogo for $2466 and up

The ONEXPLAYER X2 Mini Pro is a handheld game console with an 8.8 inch, 1920 x 1200 pixel, 30 -144 Hz AMOLED display, detachable controllers, and the most powerful graphics currently available for a device this size.

But with an AMD Strix Halo processor and at least 48GB of high-speed memory inside, it’s unsurprising that this handheld is expensive. The ONEXPLAYER X2 Mini Pro is now available […]

ONEXPLAYER X2 Mini is a Ryzen AI Max+ 395 handheld gaming PC with detachable controllers

The ONEXPLAYER X2 Mini is an upcoming handheld gaming PC with a feature list that’s too good to be… cheap.

It has an 8.8 inch OLED display with support for variable refresh rates, detachable Bluetooth controllers that let you use the system like a game console or a tablet, and a magnetic keyboard that turns the system into a sort of tiny laptop. And with an AMD Ryzen AI Max+ 395 processor, […]

ONEXStation is a Ryzen AI Max+ 395 mini PC with 128GB RAM and RBG lighting

Chinese PC maker One Netbook is better known for making handheld gaming PCs than desktop workstations. But the company is positioning the new ONEXStation as a “mini AI Workstation” thanks to its AMD Ryzen AI Max+ 395 processor and 128GB of high-speed, high-bandwidth memory.

You could also certainly use it as a gaming PC though – and it’s got RGB lighting effects to help match the gamer […]

ONEXPLAYER Super X tablet with AMD Strix Halo is priced at $1899 and up during crowdfunding

The ONEXPLAYER Super X is a tablet with a 14 inch OLED display, a detachable keyboard, and support for pressure-sensitive input. But unlike most tablets with those kinds of features, the Super X is a tablet with the kind of horsepower you’d expect from a gaming laptop or a portable AI workstation.

First introduced earlier this year, the Super X supports up to an AMD Ryzen AI Max+ 395 […]

AMD Strix Halo handheld PC comparison: AYANEO NEXT II vs GPD Win 5 vs OneXFly Apex

AMD’s Ryzen AI Max and Max+ processors, also known by the code-name “Strix Halo” are mobile chips that combine up to a 16-core, 32-thread Zen 5 CPU with up to 40 RDNA 3.5 GPU compute units. Positioned as a solution for gaming laptops and mini PCs that can function as AI workstations, the chips are also starting to show up in handheld gaming computers.

But fitting a chip this powerful into a […]

OneXFly Apex Strix Halo handheld gaming PC hits Indiegogo for $1399 and up (options include liquid cooling and spare batteries)

The OneXFly Apex is one of only a few handheld gaming PCs to offer discrete-class integrated graphics thanks to an AMD Strix Halo processor. First introduced earlier this fall, the handheld launched in China recently for around $1200 and up. Now the company behind the handheld is preparing for a global launch.

One Netbook is running a crowdfunding campaign for the OneXFly Apex with reward […]

#handheldGamingPc #oneNetbook #onexflyApex #ONEXPLAYER #strixHalo

Read more: https://liliputing.com/onexfly-apex-strix-halo-handheld-gaming-pc-to-sell-for-1399-and-up-during-crowdfunding/

The OneXFly Apex is a handheld gaming PC with an AMD Strix Halo processor featuring discrete-class graphics, an 8 inch, FHD+ 120 Hz display, and an expected $1200(ish) starting price. Crowdfunding begins Nov 9: https://x.com/OnexPlayer_/status/1984183882707878147 #ONEXPLAYER #OneNetbook #OneXflyApex #HandheldGamingPC #StrixHalo

OneXFly Apex handheld gaming PC with AMD Strix Halo launches soon for $1200 and up (in China)

The OneXFly Apex is a handheld gaming PC with an 8 inch, 1920 x 1200 pixel 120 Hz display, an AMD Strix Halo processor with discrete-class graphics, an optional liquid cooling system that allows the chip to run at up to 120 watts, and an 85 Wh external battery pack.

After teasing the upcoming handheld in September, One Netbook has launched a preview page for an upcoming crowdfunding campaign. […]

#crowdfunding #handheldGamingPc #oneNetbook #oneplayer #onexflyApex #strixHalo

Read more: https://liliputing.com/onexfly-apex-handheld-gaming-pc-with-amd-strix-halo-launches-soon-for-1200-and-up-in-china/

Lilbits: Turning a “disposable” vape into a web server, giving the Framework Laptop a mechanical keyboard, and more

Over the past few years I’ve seen plenty of DIY projects where people have used the batteries in discarded vapes to create rechargeable power banks. But BogdonTheGeek found a use for some of the other components found in a disposable vape.

It turns out some recent vapes have a PUY PY32 microcontroller featuring a 24 MHz ARM Cortex-M0+ processor, 24 KiB of flash storage, and 3KiB of static […]

#campus #campusLaptop #disposableVape #framework #frameworkLaptop #frameworkMainboard #lilbits #mechanicalKeyboard #microsoft #minisforum #minisforumMsS1Max #puyP32 #strixHalo #vape #webServer #xbox #xboxOnPc

Read more: https://liliputing.com/lilbits-turning-a-disposable-vape-into-a-web-server-giving-the-framework-laptop-a-mechanical-keyboard-and-more/

OneXPlayer Super X gaming tablet with Ryzen AI Max+ 395 is basically an Asus ROG Flow Z13 alternative

The ONEXPLAYER Super X is an upcoming tablet with a 14 inch, 2880 x 1800 pixel, 120 Hz  AMOLED display with variable refresh rate support, an AMD Ryzen AI Max+ 395 processor with discrete-class integrated graphics and the ability to run at up to a 120W TDP.

In a lot of ways the tablet is similar the the Asus ROG Flow Z13 that launched earlier this year, but ONEXPLAYER is trying to position […]

#2In1 #gamingTablet #oneNetbook #onexplayer #onexplayerSuperX #strixHalo

Read more: https://liliputing.com/onexplayer-super-x-gaming-tablet-with-ryzen-ai-max-395-is-basically-an-asus-rog-flow-z13-alternative/

GPD Win 5 handheld gaming PC specs & performance preview: Strix Halo processor, 7 inch display, and no keyboard

This week GPD revealed that its next handheld gaming PC would be powered by an AMD Strix Halo processor, bringing discrete-class graphics to a handheld PC for the first time. But at the time the company didn’t share many other details about the upcoming GPD Win 5, which left questions about battery life and overall usability.

Now The Phawx has published a video revealing more features. The GPD […]

#gpdWin5 #handheldGamingPc #strixHalo #win5

Read more: https://liliputing.com/gpd-win-5-handheld-gaming-pc-specs-revealed-strix-halo-processor-7-inch-display-and-no-keyboard/