Realtek RTL8159 10GbE to USB 3.2 adapters sell for about $55 and up

Realtek unveiled low-cost, low-power 10GbE RTL8127 (PCIe) and RTL8159 (USB 3.0) controllers at Computex 2025. Since then, we noted that RTL8127-based M.2 and PCIe 10GbE cards had already launched for as low as $35. However, a new X post by Jeff Geerling reminded me it was time to check on those RTL8159 10GbE to USB 3.2 adapters. So let's do that now, starting with the Wisdpi WP-UT9 mentioned in the post/tweet. Wisdpi WP-UT9 WisdPi WP-UT9 specifications: Chipset - Realtek RTL8159 Ethernet - Up to 10GbE RJ45 port; backward compatible with 5 Gpbs, 2.5 Gbps, 1Gbps, and 10/100Mbps Ethernet USB - USB 3.2 Gen2 Type-C port; backward compatible, but slower ports will impact performance; USB hubs/docks and long cables are not recommended either Misc - 2x LEDs - Amber (Link) / Green (Activity) Power Supply - Via USB-C port Dimensions - 59 x 29 x 13 mm Weight - 58.8 grams

CNX Software - Embedded Systems News

📈 GPT-5.4 is out with ~33% fewer factual errors vs GPT-5.2 and better coding scores.

Steady gains. No single 'AGI moment'.

The real story is how fast incremental progress compounds.

#GPT5 #OpenAI #LLM #Coding #Benchmarks

🧠 AI jumped from near-zero to 37% on an 'unbreakable' expert exam — in just 14 months.

Humans still lead. But the gap is closing faster than anyone expected.

What happens when it hits 80%?

#AI #Benchmarks #Research #AGI #ML

Nicolò Boschi (@nicoloboschi)

Apify의 스크래핑 기능이 AI 에이전트에게 더 풍부한 컨텍스트를 제공할 수 있다고 언급하며, 오픈소스 장기 기억(long-term memory)을 추가하려면 관련 벤치마크를 확인해보라고 제안한다. AI 에이전트, 데이터 수집, 메모리 평가에 관심 있는 개발자에게 유용한 참고 트윗이다.

https://x.com/nicoloboschi/status/2038954037110820992

#apify #aiagents #webscraping #longtermmemory #benchmarks

Nicolò Boschi (@nicoloboschi) on X

@moritzkremb Apify's scraping capabilities can significantly enhance the context available to AI agents. To add open source long term memory, it is worth checking the benchmarks. https://t.co/c5F9bMmgdi

X (formerly Twitter)
How bad will Europe’s energy crisis get?

Germany’s Friedrich Merz warns the economic fallout from the war in Iran is on track to rival that of the Covid pandemic or the Russian invasion of Ukraine.

POLITICO
Comprehensive C++ Hashmap Benchmarks 2022

Where I've spent way too much time creating benchmarks of C++ hashmaps

Martin Leitner-Ankerl

📊 ARC-AGI-3 is out — and it's humbling today's best models.

Frontier AI still can't crack flexible, general reasoning at human level.

The gap is real. AGI hype needs a reality check.

#AGI #Benchmarks #AI #Research #ARC

TerraMaster F2-425 Plus NAS review – Part 2: Configuration, benchmarks, and AI-enhanced media storage

I received the TerraMaster F2-425 Plus 3+2 Hybrid NAS for review last month, and after checking out the hardware in the first part of the review, I've finally had time to test the Intel N150 NAS. After installing two 4TB SATA drives and an M.2 NVMe SSD, I'll report my experience setting up the system with the TNAS Android app, before running some benchmarks, and testing features like photo backup with AI search capabilities. Hard drive installation I already had an old, but little-used, 4TB HGST SATA drive, and I bought a "new" 4TB SATA drive online for a pretty good deal (2979 THB or a little over $90 US). It turns out the HPE MB4000GVY2K drive I got was refurbished, having been manufactured in 2017. However, it's an enterprise-grade drive, and the TNAS app reports it has been used "only" for 2,517 hours, so I don't feel too bad

CNX Software - Embedded Systems News

Matlab Alternatives 2026: Benchmarks, GPU, Browser and Compatibility Compared

https://runmat.com/blog/free-matlab-alternatives

#HackerNews #MatlabAlternatives #2026 #Benchmarks #GPU #BrowserCompatibility

MATLAB Alternatives 2026: Benchmarks & Compatibility Compared

We benchmarked 4 MATLAB alternatives on real engineering workloads — loop performance, GPU acceleration, and .m file compatibility. One runs existing scripts with zero rewrites.

OpenAI führt mit Model Spec Evals einen neuen Benchmark zur Messung der Regeltreue ein. Dabei schneidet das ältere GPT-5 Thinking mit 89 Prozent messbar besser ab als das aktuelle GPT-5.4 Thinking. Modelle mit Reasoning-Fähigkeiten erweisen sich bei der Einhaltung von 225 Verhaltensregeln als robuster gegenüber kompakten Architekturen.

#OpenAI #GPT5 #LLM #Benchmarks #News
https://www.all-ai.de/news/news26/gpt4o-test-gpt5-4

Warum das beste GPT-Modell nicht das neueste ist

Die Auswertung hunderter Test-Prompts offenbart erstaunliche Leistungsunterschiede. Eine bestimmte Architektur setzt sich klar von der Konkurrenz ab.

All-AI.de