luokai
banner
luok.ai
luokai
@luok.ai
For more AI&Tech content, check here www.luok.ai
🍎Apple Die Hard Fan| 苹果骨灰粉
🤖GenAI Observer | GenAI观察者
👨🏻‍🎤Cutting Edge Tech Enthusiast | 科技爱好者
Space DJ turns genre embeddings into a playable galaxy—pilot a ship, the music follows. 🚀

Key stats
768→128 PCA compression; 3D UMAP projection; three.js rendering; autopilot drift; high‑dim neighbors surfacing hidden similarities.
November 11, 2025 at 3:03 PM
ALLWEONE drops an open-source rival to Gamma—AI-crafted slides, your way. ⚙️✨

It builds full decks from a topic, lets you edit outlines, pick themes, and watch slides render in real time. Choose pro or casual style, swap image models, and present inside the app.
November 11, 2025 at 2:38 PM
Meta just dropped a universal speech engine—Omnilingual ASR—built to hear the world.

Key stats
1600+ langs - 500 first-time served - 300M–7B params - w2v 2.0 (7B) backbone.
November 11, 2025 at 2:35 PM
Social platforms breed polarization by design—can “quick fixes” really help?

A new study tests prosocial interventions using Generative Social Simulation (LLMs inside agent-based models).

www.threads.com/@luokai/post...
luokai (@luokai) on Threads
Social platforms breed polarization by design—can “quick fixes” really help? A new study tests prosocial interventions using Generative Social Simulation (LLMs inside agent-based models). Even a mini...
www.threads.com
November 11, 2025 at 2:32 PM
WhatsApp is now available on Apple Watch.
November 6, 2025 at 3:21 AM
Gemini Deep Research just got a brain upgrade: it now taps your Google Workspace.

Pull context straight from Gmail, Drive (Docs/Slides/Sheets/PDFs) and Google Chat to auto-build richer reports.
November 6, 2025 at 3:12 AM
AI isn’t replacing craft—it’s supercharging the 3D pipeline.

A French studio delivered a 2026 Olympics ad with a classic CG workflow enhanced by AI. Roles even list “AI artists,” showing AI as a legit production discipline, not a gimmick—more iteration, faster previz, sharper polish.
November 6, 2025 at 1:52 AM
A canvas that looks like paint—but updates with your voice.

Fraimic is a color E Ink “smart canvas” that swaps art via voice, phone, or web—no app needed, no internet required. It fits standard frames, runs on a multi‑year battery, and uses Spectra 6 so images read like real pigment on canvas.
November 6, 2025 at 1:45 AM
Toyota just demoed a personal mobility chair that literally walks—up stairs, sideways, and into a car.

Meet “walk me”: a four-legged, stair-climbing, height-adjustable mobility device aimed at seniors and people with limited mobility.
November 5, 2025 at 4:53 PM
Gemini’s Canvas goes from blank slide to polished deck—just say “create a presentation.”

Key stats
Launch highlight: single‑prompt slide generation, editable text in-slide, image regeneration per slide (requested), file attach limits (requested), regional rollout ongoing.
November 5, 2025 at 4:44 PM
SkyReels V3 just turned “ideas → full productions” into a one-canvas workflow. ⚡

Key stats
1 Super Agent + 28 Expert Agents; 32 camera moves; ultra‑long scenes; prompt‑controlled dialogs.
November 5, 2025 at 4:43 PM
Sora just dropped character cameos—your recurring cast, now reusable across videos. ⚡

Key stats
Now in the Sora app; supports cameos + stitching + leaderboard; rollout appears region‑limited per user reports.
October 31, 2025 at 4:58 PM
Dia just shipped three small upgrades with outsized impact. ⚡

Meet Focus Mode, Custom Shortcuts, and Google Meet PiP. Designed for fewer distractions, faster actions, and seamless calls while you work. Techy, tidy, and practical—exactly what a modern AI browser should feel like.
October 31, 2025 at 4:55 PM
Affinity just flipped the table: one app to design, photo‑edit, and layout — for free.

Key stats
Free download on macOS & Windows; existing V1/V2 users report bonus fonts.
October 31, 2025 at 4:53 PM
Cursor 2.0 flips coding: agents first, files second. 🎯

Key stats
Parallel agents, multi‑model comparison for harder tasks, fast change review, built‑in browser tool for automated testing.
October 31, 2025 at 4:50 PM
Google Labs drops Pomelli: an AI that turns your website into on‑brand campaigns—fast. ⚡

Key stats
Availability: US, Canada, Australia, New Zealand. Launch traction: traffic surge reported; service restored. Entry: website input → campaign generation.
October 31, 2025 at 4:47 PM
Alibaba has moved from its open-source Qwen model to AI glasses—declaring war on Meta in hardware.👓

Key stats
Dual hot-swap “leg battery” for 24h use vs Meta’s ~6–8h. 3K capture with AI upscaling to 4K, prescription-ready design.
October 29, 2025 at 3:20 PM
Hailuo 2.3 just leveled up AI video—realism that actually feels… real. 🚀

Key stats
Two modes: Hailuo 2.3 (cinematic fidelity) vs Hailuo 2.3 Fast (speed/cost). 4 free videos daily for everyone.

4 free videos daily → hailuoai.video
October 29, 2025 at 3:14 PM
Claude just went finance-native: live market data, Excel beta, and agent skills that actually do the work. 🚀
October 29, 2025 at 3:11 PM
A linnet’s song, turned into 3D islands you can “see.” 🐦✨

Lucioarese visualizes bird syllables by mapping acoustic proximity into point-cloud clusters, revealing motifs and textures in Thibault Noirot’s original recording.
October 29, 2025 at 3:08 PM
Google is leveling up Earth AI with Gemini-powered geospatial reasoning—faster insights, bigger reach. 🚀

Key stats
Flood forecasting now covers 2B+ people; LA crisis alerts reached 15M during 2025 wildfires; new models for floods, wildfires, cyclones, air quality.

youtu.be/kpeI9dbkWho?...
Geospatial Reasoning: Demonstration application (October 2025)
YouTube video by Google Research
youtu.be
October 29, 2025 at 1:07 AM
Meet LTX-2—an open-source challenger to commercial AI video suites, built for creators and pipelines. 😎

Key stats
4K native, 50 fps, 10s+ sequences; API-first; consumer GPU target (5‑series realistic post‑optimization).

You can test LTX‑2 in the API Playground today. app.ltx.studio
October 29, 2025 at 1:03 AM
Vidu Q2 just leveled up reference-to-video: multi-image roles, smoother cams, lower cost. ⚡

Key stats
2 images → full 3:54 demo; noted smoother camera; priced “very low” vs peers; solves Sora2’s real-face upload gap.

Try it now: service.vidu.com/vidu/v1/shar...
October 22, 2025 at 11:24 PM
Galaxy XR brings multimodal AI into spatial computing—turning maps, media, and memories into live, interactive scenes. 🤖

youtu.be/r0AyzvfTRLo?...
Introducing Galaxy XR | Samsung
YouTube video by Samsung
youtu.be
October 22, 2025 at 11:22 PM
ChatGPT gets a native browser: Atlas brings agent mode + memory to the web.

Key stats
Launch: worldwide on macOS (Free/Plus/Pro/Go; Business beta; Enterprise/Edu if enabled).
Agent mode: preview for Plus/Pro/Business.
Controls: incognito, per‑site visibility, optional training opt‑in.
October 21, 2025 at 11:33 PM