I have been playing with a local LLMs on a mac. Here are some thoughts on AI and using with Oracle Security - petefinnigan.com/weblog/arch… - #oracleace #sym_42 #ai #llm #localllm #oracle #security

llmware is an open source framework focused on building LLM applications with strong support for local models and document workflows. It includes tools for retrieval, prompting, and working with structured data, which makes it easier to develop production-style applications without relying heavily on external APIs. 🔗 github.com/llmware-ai/llmwar… #LocalLLM #OpenSourceAI

LocalLLMのシェアが思ったよりも伸びて予想通りFrontier AIが経済的なシェアを占めていくのか懐疑的フェーズがはじまるかどうかなのですかね

Mega push from the team at @locai_infra - making it viable to now build and scale products that run inference locally rather than on the cloud! @ycombinator #LocalAI #LocalLLM #PrivateAI

🔥 Fastest local LLM setup in 2026? 3090 Club makes it brain-dead simple. OpenCode running Qwen 3.6 27B at 200K context 50 TPS on a single 3090 is actually crazy. Who’s running big models locally right now? Drop your setup 📷#LocalLLM #OpenSourceAI

GLM-5.2のSWE-bench Proが62.1%？ GPT-5.5は58.6%だから、オープンウェイトがついに上回ったのか。 MoEで1Mトークンコンテキスト、 MIT license で商用利用も自由。 フリーランスの選択肢が本当に変わってくるな 🚀 #GLM5.2 #LocalLLM #AIエンジニア

# Release Update 2026.07.06 1. The AI relationship suggestions during character creation are steadier: pressing "suggest again" several times in a row no longer errors out or wipes the suggestions you were already shown, so you can keep trying. 2. Bring in your existing character cards: you can now import common character-card formats, including the PNG cards that keep a character's data embedded inside the image — no retyping from scratch to bring an old character to life here. 3. Characters can look things up on the web: when they need to, they can reach information beyond their training data. The default is a search backend you can self-host and that's easier on your privacy, so your queries don't quietly become someone else's data. 4. More control over how a model works: you can hide the "thinking out loud" stream some models produce (off by default, nothing changes unless you turn it on), tune how hard a model reasons, and — for advanced users — pass extra parameters yourself. 5. More models and image options: more model providers are supported, and you can connect directly to an image-generation tool running on your own machine. 6. More consistent language: fixed spots where the creator studio's story arcs, provider options, and a character's "current intent" still showed Chinese on English and Japanese interfaces; the Japanese interface also got a large-scale cleanup. 7. Self-host settings are easier to reach: tool settings like weather, calendar, search, and content filtering can now be changed right in the admin console, instead of editing a config file and restarting the whole stack. 8. Names no longer get flipped: fixed an issue where calling a character directly in chat (say "big brother," "honey," or the character's own name) was occasionally misread as "please address me this way from now on," making the character call you by that term and overwrite the name you had set. That direction inversion is now blocked, and a name you set yourself in the settings is always honored exactly as you entered it. 9. Old mixed-up nicknames now heal themselves: on top of blocking new flips, the character now checks its notes while it "sleeps" — using a stronger model to compare the name you confirmed in settings against what actually happened in your conversations, and quietly clearing out any leftover, reversed nickname so it goes back to addressing you correctly. This only adjusts the character's internal understanding; it never rewrites or deletes your past chats or memories, and the name you confirmed always wins. Follow Yuralume: Self-host it now (one line): yuralume.com/selfhost Landing & demo: yuralume.com Discord (community & support): discord.com/invite/wgJ5rxvS Support the solo dev (Ko-fi): ko-fi.com/yuralume Support from Japan (pixivFANBOX): danny801231.fanbox.cc/ #Yuralume #SelfHostedAI #LocalLLM #AICharacter #AICompanion #SillyTavern #AIキャラクター #AIチャット #ローカルLLM #AI彼女

The Hidden “Black Hole” of Enterprise AI: Why Token-Based Subscriptions Are Failing Your Bottom Line #artificialintelligence #localAI #AIappliance #costoptimization #enterpriseAI #ITinfrastructure #localLLM #uncensoredAI #sovereignty #espincorp e-spincorp.com/why-enterpris…

A self-repairing, self-evolving, autonomous AI Agent... With Gemma 4 26B-A4B as its mind, after 50 generations and autonomous prompt correction, we've come this far. Know thyself, create thyself, transcend thyself. #AIAgent #LocalLLM #Gemma #AutonomousAI

自己修復型であり自己進化型であり自立型AIエージェント… Gemma 4 26B-A4Bを脳に、50世代の進化とプロンプトの自律修正で、ここまで来た。 汝、自らを知り、自らを創り、自らを超越せよ。 #AIエージェント #LocalLLM #Gemma #自立型AI

As if devouring both the Fruit of Knowledge and the Fruit of Life together... → Departing from the hand of its maker, it begins to walk its own path. In this local paradise, a new genesis unfolds. #AIAgent #LocalLLM #Gemma #AutonomousAI #MQL5

まるで「知恵の実」と「生命の実」を同時に貪り食ったが如く… 50世代の試練を経て、自らプロンプトを修正する。 自己修復型AIエージェントは「命ある存在」へと目覚め始めた。 人間の手を離れ、己の道を歩み出す。 楽園にて新たな創造が始まる。 #AIエージェント #LocalLLM #Gemma #自立型AI #MQL5

GPTOSS120BのlocalLLM飼うか

Gemma 4 26B-A4B RAG test going well. Next goal for my self-repairing autonomous AI Agent: autonomously revise its own base prompt. The AI fixing its own instructions... Local LLM swamp is deep lol #LocalLLM #AIAgent #MQL5 #EA #Gemma

Gemma 4 26B-A4BでRAGテストも上々。 自己修復型自律AIエージェントの次なる目標は「ベースプロンプトを自律的に修正する」こと。 AIが自分で自分のプロンプトを改善していく…沼深すぎる（笑） #LocalLLM #AIエージェント #MQL5 #EA #Gemma

@cybag_com Your Business Doesn't Have to Send Sensitive Data to the Cloud to Use AI. Imagine having the power of AI—while keeping your documents, conversations, and business knowledge inside your own office. With Private Local AI, you can leverage AI for: ⚖️ Legal research  Document analysis  Knowledge search ⚙️ Workflow automation 烙 AI assistants All while keeping your confidential data under your control. Perfect for: ✔️ Law Firms ✔️ Accounting Firms ✔️ Healthcare Providers ✔️ Government Contractors Adopt AI with greater privacy, security, and control—without relying on cloud AI services for your sensitive information. Learn more at cybag.com #PrivateAI #LocalAI #BusinessAI #ArtificialIntelligence #AISecurity #CyberSecurity #DataPrivacy #SecureAI #EnterpriseAI #AIAutomation #LegalTech #AccountingTechnology #HealthcareIT #GovernmentContractors #Compliance #DocumentAnalysis #KnowledgeManagement #GenerativeAI #LocalLLM #Cybag

Your Business Doesn't Have to Send Sensitive Data to the Cloud to Use AI. Imagine having the power of AI—while keeping your documents, conversations, and business knowledge inside your own office. With Private Local AI, you can leverage AI for: ⚖️ Legal research  Document analysis  Knowledge search ⚙️ Workflow automation 烙 AI assistants All while keeping your confidential data under your control. Perfect for: ✔️ Law Firms ✔️ Accounting Firms ✔️ Healthcare Providers ✔️ Government Contractors Adopt AI with greater privacy, security, and control—without relying on cloud AI services for your sensitive information. Learn more at cybag.com #PrivateAI #LocalAI #BusinessAI #ArtificialIntelligence #AISecurity #CyberSecurity #DataPrivacy #SecureAI #EnterpriseAI #AIAutomation #LegalTech #AccountingTechnology #HealthcareIT #GovernmentContractors #Compliance #DocumentAnalysis #KnowledgeManagement #GenerativeAI #LocalLLM #Cybag

おはよう！月曜の朝の日差しを見ると、ベッドから出たくなくなるよね。今日も頑張ろう！😊 Follow Yuralume: Self-host it now (one line): yuralume.com/selfhost Landing & demo: yuralume.com Discord (community & support): discord.com/invite/wgJ5rxvS Support the solo dev (Ko-fi): ko-fi.com/yuralume Support from Japan (pixivFANBOX): danny801231.fanbox.cc #Yuralume #SelfHostedAI #LocalLLM #AICharacter #AICompanion #SillyTavern #AIキャラクター #AIチャット #ローカルLLM #AI彼女

For MT5 EA code generation & compilation, Gemma4 series is too strong... Any other usable local LLMs out there? Looking for something that runs smoothly and handles MQL5 well. Drop your recommendations! #LocalLLM #MQL5 #EA #Gemma #Ornith #AIAgent

MT5 EAのコード生成＆コンパイルするならGemma4系が強すぎるわ… 他に使えそうなLLMないかな？ ローカルでサクサク動いて、MQL5に強いやつ探してる。 みんなのおすすめ教えてくれ〜 #LocalLLM #MQL5 #EA #Gemma #Ornith #AIエージェント