China Proposes Export Bans on 40 AI, Chip and Solar Techs
..................
https://nile1.com/china-proposes-export-bans-on-40-ai-chip-and-solar-techs/
..................
#2TrillionParameters #aiExportControls #chinaExportBan #kimiK3 #ministryOfCommerce #polysiliconExportRestrictions #qwen3.8-max #

[JP] LLMのNO-THINK強化学習による性能向上、実は元のTHINKモードが「漏洩」して再活性化しただけ？
[EN] Performance Boost in LLMs through NO-THINK Reinforcement Learning: Could It Be Just the Original THIN…

https://ai-minor.com/blog/en/2026-10-16-1792135164087-the_thinking_leakage_in_reinforcement_learning_for

#LLM #強化学習 #Qwen3 #AI #Tech

NanoJev can juggle changing candidate lists for LLM routing, but picking the right model is a whole other game. See how Qwen3‑0.6B, Kev & Verdict stack up in the new decision‑model showdown. #NanoJev #Qwen3 #DecisionModels

🔗 aidailypost.com/news/nanojev...

Yep, that works.
#ai8 #ai #aiart #aiartwork #aigenerated #aicommunity #aiartcommunity #aitraining #digitalart #creativeai #synthart #generativeart #Strata #LLM #Qwen3.8-flash-next #Qwen3.8

#Clefomni adds #audiototext and #videototext and #imageinputs via #Qwen3 Omni-30B-A3B-Instruct. #Clefflash is #cheaper than #Jev at $0.038 per M tokens with a 24k context window; #Clef is #faster via #SGLang. https://blog.cloudfl... #tech #news #ainews

Zeit für ein vorläufiges Résumé: #Qwen3 4B über Ollama: für inhaltliche Fragen an das #Paperless-Archiv. Qwen3 1.7B über Ollama: für einfachere Aufgaben, wenn die Geschwindigkeit wichtiger ist. #Vulkan: vorerst als experimentelle Option, nicht als produktiver Paperless-Backend-Dienst geeignet.

📏 262,144 tokens of native context, validated up to 1,048,576. 40 of 50 layers use 512-token sliding-window attention, every 5th layer attends to everything. RULER at 1M tokens: 63.2 vs 57.5 for #Qwen3.5 35B-A3B

#freegame #textbased #dnd5e #roleplaying #deepseek #qwen3.8 #stepfun #step-5-preview #tts #qwentts

2×V100 で Qwen3.8-Flash-Next を 256K コンテキスト・約 88 tok/s で動かす：llama.cpp v0.5.0 に加えた最適化

#Qwen3.8-Flash-Next #LocalLLM #llama.cpp

Qwen 3.6 35B A3B, running locally on my gaming PC: RTX 5060 (8GB VRAM), Ryzen 7 7700, 32GB RAM.

Full 6:19 run. No cuts or speed-ups.

Temps are stable with my llama config. AAA games push this machine hotter than the local LLM.

#Qwen3 #LocalLLM #RTX5060

Amazon SageMaker AIを活用したQwen3-TTSによるリアルタイム・パーソナライズ音声合成の展開

Qwen3-TTSとSageMakerで実現するリアルタイム音声合成の技術と展開

#テキスト読み上げ(TTS) #Qwen3-TTS #AmazonSageMakerAI #リアルタイムAI #音声クローニング

Small footprint, big benchmarks: Qwen3.8 27B is now on Scaleway Generative APIs 🚀 Best open-weight performer in its size class on most benchmarks (Aug 2026). Plug it into your favorite agentic tools or IDE. #Qwen3 #GenerativeAI #OpenWeight #LLM

Alibaba's Qwen3.8-Omni-Flash claims 98% cheaper audio than its predecessor. The catch: that's a per-hour figure built from a 2-minute sample x30, and the audio+video version of the claim…

https://rtfclmgzn.com/?utm_source=bluesky&utm_medium=social&utm_campaign=autopost#
#Alibaba #Qwen3 #OmniFlash

ハーネス（OpenCode）＋ローカルLLM ７モデル比較で Qwen3.8-Flash-Next が満点！

#ハーネスエンジニアリング #opencode #Qwen3.8-Flash-Next #LLM #ベンチマーク

Just saw AIPerf’s latest benchmark—Qwen3‑0.6B crushing LLM inference speeds. If you care about real‑world AI performance, this dive is a must‑read. Curious how it stacks? Check it out! #AIPerf #Qwen3 #LLMInference

🔗 aidailypost.com/news/aiperf-...

[JP] 27Bモデルが5.9GBに！9倍軽量化した「Ternary Bonsai 2 27B」がローカルAIの常識を変えるサメ！
[EN] The 27B Model Shrinks to 5.9GB! The Lightweight "Ternary Bonsai 2 27B" is Set to Revolut…

https://ai-minor.com/blog/en/2026-09-18-1789685583672-bonsai_2_27b__near_lossless_compression_in_a_9x_sm

#ローカルLLM #量子化 #Qwen3.8 #AI #Tech

#Dezgo, which I use daily, is a fantastic and mostly free #AIArt and #AIVideo model collection. They have just announced three new image models on the paid side, including the uncensored model #Qwen3. The other two are broken puritan trash. dezgo.com/app/image2im...

Qwen3.8-27B-Uncensored from OrcaRouter Achieves Top Rank in Security Model Comparison #Japan #Chiyoda #FlashLabs #OrcaRouter #Qwen3.8

「Qwen3.8-27B-Uncensored」がセキュリティ研究で1位獲得、品質を証明する朗報 #AIモデル #OrcaRouter #Qwen3.8-27B

FlashLabsが提供するAIモデル「Qwen3.8-27B-Uncensored」が、Abliterliticsによる比較検証で総合1位を獲得。この成功の背景と性能を詳しく解説します。