Zlio
Everything in AI and robotics, on one page.
Updated 15:33 UTC · Next update 16:00 UTC
Zlio Daily, the newsletter, is coming soon. The day's essential AI and robotics news in your inbox. Leave your email and we'll tell you when it launches.
One email when we launch. No spam, unsubscribe anytime. Privacy

Latest news

Unite.AI · 3h #

Translated releases Lara Mini 45-language research model

Translated released Lara Mini, a 7.2B-parameter mixture-of-experts model that activates 600M parameters per token. Trained from scratch on 6 trillion tokens across English and 44 other languages, including all 24 official EU languages.

Translated
Hugging Face Blog · 6h · 3 sources #

llama.cpp hits v1.0.0, setting API stability rules

Why it matters: A 1.0 milestone for the most widely used local LLM runtime, with stability guarantees for tools.

llama.cpp, the open-source project for running LLMs locally on your own hardware, reached version 1.0.0, establishing API stability rules for the local AI ecosystem.

ggml-orgOpenCodeOpen weightsFree
CX Today · 6h · 22 sources #

Interactive Brokers offers European clients access to Anthropic IPO

Why it matters: Retail access to a mega AI IPO broadens who can bet on frontier labs.

Interactive Brokers is giving eligible European clients access to the Anthropic PBC initial public offering, as analysts weigh a reported $1.4 trillion valuation and rising enterprise AI bills.

Anthropic
Product Hunt · 7h #

Burnstate launches Mac menu bar app for AI coding limits

Burnstate is a native Mac menu bar app that shows Claude Code, Codex, Cursor and Vercel AI Gateway limits, estimates when you'll run out, answers Claude Code permission prompts, and tracks session tokens and cost.

Launch · ▲ 80 on Product Hunt
BurnstateAppPaid
RIABiz · 10h #

Schwab launches MCP servers, moves to agentic AI

Why it matters: A major brokerage adopting agentic MCP servers signals agent access to financial accounts.

Charles Schwab surprised analysts with fast launch of its own MCP servers, moving from read-only to agentic AI that can complete tasks. RIAs are excluded for now; Anthropic's exclusive LLM role may end.

Charles SchwabAPIOpen finance
DigiTimes Asia · 12h #

SpaceX-Tesla merger talks advance toward $3.4 trillion deal

Why it matters: A record-scale tie-up would reshape one of AI's biggest corporate players.

SpaceX and Tesla are advancing merger talks that could create a US$3.4 trillion conglomerate spanning EVs, aerospace and AI, with an announcement possibly as early as December 2026, though China ties complicate approval.

SpaceX
Pandaily · 13h #

Huawei Cloud to launch Ascend-native CloudRobo-WM1.2 Nov 16

Why it matters: A world model built natively on Ascend chips signals Huawei's push into physical AI.

Huawei Cloud will unveil CloudRobo-WM1.2, billed as the first Ascend-native world model from its Industry AI Dream Factory, at an online launch on Nov 16; specs are not yet public.

Huawei CloudWaitlist
DigiTimes Asia · 13h · 7 sources #

Asus unveils Gallop humanoid robot and posts record Q3 revenue

Why it matters: Asus joins the humanoid race with a factory-proven robot for electronics assembly.

Asus announced Gallop, its first humanoid robot, which learns to pack tiny motherboard parts in a factory. Asus also reported record Q3 brand revenue of NT$258.788B (US$7.86B), up 36% year-on-year, on AI server shipments.

Asus
The Economic Times · 20h Widely covered · 25 sources #

Warby Parker, Samsung launch Gemini-powered AI glasses

Why it matters: A major consumer AI wearable launch that could set the template for Gemini-powered smart glasses.

Samsung unveiled 41g AI glasses built with Google Gemini, launching with Warby Parker and other eyewear brands; privacy and bystander concerns remain amid the rollout.

SamsungClosedOn sale
TechCrunch · 23h · 6 sources #

SoftBank takes $3.7B charge on failed AI gas turbine bet

Why it matters: SoftBank's costly turbine over-ordering shows how power supply is the bottleneck for AI data centers.

SoftBank lost over $3 billion pre-ordering natural gas turbines to power AI data centers, driving a $3.7bn charge; separately Anthropic's Dario Amodei predicted AI could cure major diseases within a decade.

SoftBankDatacenters
The Elec · 1d #

AMD launches 6th-gen EPYC Venice server CPUs

Why it matters: AMD's new server CPUs are a direct challenge to Nvidia in AI datacenter compute.

AMD launched its Zen 6-based sixth-generation EPYC 'Venice' server CPUs, aiming to gain share with better price-performance than rivals; AMD claims 3.75 times Nvidia performance.

AMDChipsDatacenters
The Robot Report · 1d · 3 sources #

Runway releases Praxis-1 open-weights robot world action model

Why it matters: An open-weights action model from a video-AI leader could accelerate robot world-model research.

Praxis-1 is a world action model that lets robots learn actions from video-style world modeling; Runway released it with open weights under Apache 2.0, developed with partners Noble Machines, Standard Bots and Ultra.

RunwayOpenOpen weightsCode
Hugging Face · Hugging Face · 1h #

Higurashi-ai releases Haru-4B full-duplex spoken dialogue model

Haru-4B is a 4B-parameter audio-to-audio foundation model that listens and speaks simultaneously, like a phone call, at 7.5 Hz frame rate with ~133 ms latency. Apache-2.0 weights on Hugging Face; research base model.

Model · audio-to-audio · apache-2.0 · ♥ 101 · 165 downloads
higurashi-aiOpenOpen weightsFree
Dev tool · GitHub · 1h #

Hugging Face Transformers v5.23.0 adds Inkling 2 support

Transformers, the Hugging Face library for defining and running ML models across text, vision and audio, released v5.23.0, adding Inkling 2 Small, a multimodal model from Thinking Machines Lab.

Release v5.23.0 · huggingface/transformers
Hugging FaceOpenCodeFree
Hugging Face · Hugging Face · 3h #

Edge0-145B-NVFP4 quantized model lands on Hugging Face

A quantized NVFP4 build of Edge0-145B, a 145B frontier-grade on-device text model built on Qwen3.8-Flash-Next, is available for download on Hugging Face.

Model · text-generation · other · ♥ 101 · 14.1k downloads
Edge0OpenOpen weightsFree
Hugging Face · Hugging Face · 3h #

Browser demo Ava offers hands-free voice chat with 3D avatar

Ava is a Hugging Face Space demo for talking face to face with LiquidAI's LFM3-Omni-2B through a 3D lip-synced avatar. It runs fully in the browser on WebGPU with no backend, so voice never leaves the device.

Space · static · ♥ 100
Liquid AIOpenFreeOpen weights
Dev tool · GitHub · 5h #

ONNX Runtime 1.32.0 released with hybrid MoE offloading

ONNX Runtime is a cross-platform engine for running and training ML models. Version 1.32.0 adds opset 28 support, CPU/GPU hybrid MoE offloading, new QMoE kernels and faster session init.

Release v1.32.0 · microsoft/onnxruntime
MicrosoftOpenCodeFree
Dev tool · Hugging Face Blog · 6h · 3 sources #

llama.cpp hits v1.0.0, setting API stability rules

Why it matters: A 1.0 milestone for the most widely used local LLM runtime, with stability guarantees for tools.

llama.cpp, the open-source project for running LLMs locally on your own hardware, reached version 1.0.0, establishing API stability rules for the local AI ecosystem.

ggml-orgOpenCodeOpen weightsFree
App · Product Hunt · 7h #

GAtrends launches keyword spike alerts for AI assistants

GAtrends is a trend radar for chosen keywords that watches seven sources, learns each keyword's normal level and sends Telegram alerts on spikes, with a dashboard, API and Claude/ChatGPT integration.

Launch · ▲ 76 on Product Hunt
GAtrendsAppAPIFreePaid
Dev tool · Product Hunt · 7h #

Burnstate launches Mac menu bar app for AI coding limits

Burnstate is a native Mac menu bar app that shows Claude Code, Codex, Cursor and Vercel AI Gateway limits, estimates when you'll run out, answers Claude Code permission prompts, and tracks session tokens and cost.

Launch · ▲ 80 on Product Hunt
BurnstateAppPaid
Hugging Face · Hugging Face · 7h #

Nvidia releases Human-1M physical AI human sensor dataset

Why it matters: Large human-motion dataset could give robot models the scale that text gave LLMs.

A dataset of sensorized human data — egocentric video, dense language annotations, head and hand pose — meant to connect perception, action and intent for training physical AI. Hosted on Hugging Face under CC-BY-4.0.

Dataset · n<1K · cc-by-4.0 · ♥ 50
NvidiaOpenOpen weights
App · Pi · 8h #

Pi app updates to version 26.1109.0 for iPhone

Pi, a personal AI assistant app for iPhone and iPad, released version 26.1109.0 with bug fixes and performance improvements.

Version 26.1109.0 · iPhone / iPad
InflectionClosedAppFree
All app versions & downloads → All dev tool versions →
Older news → RSS →

Shipping now: what you can use today

On Hugging Face → rss

Hugging Face · 1h #

Higurashi-ai releases Haru-4B full-duplex spoken dialogue model

Haru-4B is a 4B-parameter audio-to-audio foundation model that listens and speaks simultaneously, like a phone call, at 7.5 Hz frame rate with ~133 ms latency. Apache-2.0 weights on Hugging Face; research base model.

Model · audio-to-audio · apache-2.0 · ♥ 101 · 165 downloads
higurashi-aiOpenOpen weightsFree
Hugging Face · 7h #

Nvidia releases Human-1M physical AI human sensor dataset

A dataset of sensorized human data — egocentric video, dense language annotations, head and hand pose — meant to connect perception, action and intent for training physical AI. Hosted on Hugging Face under CC-BY-4.0.

Dataset · n<1K · cc-by-4.0 · ♥ 50
NvidiaOpenOpen weights
Hugging Face · 1d #

Pika Labs demos Pika H3 Flex video-audio generation

Pika H3 Flex is a six-step post-training of MiniMax-H3 that generates video with audio, using two rank-128 LoRA adapters for text/image-to-video and reference-to-video. A Hugging Face Gradio Space offers the demo; code is released.

Space · gradio · ♥ 100
Pika LabsOpenOpen weightsCode
HF Papers · 1d #

Ropedia posts Xperience 2 physical AI dataset paper

Xperience 2 is a paper describing a large-scale dataset and environment suite for scaling human demonstration data and interactive environments used to train robots on long-horizon, contact-rich tasks.

Paper · ▲ 20 upvotes
Ropedia
More On Hugging Face →

On GitHub → rss

GitHub · 1h #

Hugging Face Transformers v5.23.0 adds Inkling 2 support

Transformers, the Hugging Face library for defining and running ML models across text, vision and audio, released v5.23.0, adding Inkling 2 Small, a multimodal model from Thinking Machines Lab.

Release v5.23.0 · huggingface/transformers
Hugging FaceOpenCodeFree
GitHub · 5h #

ONNX Runtime 1.32.0 released with hybrid MoE offloading

ONNX Runtime is a cross-platform engine for running and training ML models. Version 1.32.0 adds opset 28 support, CPU/GPU hybrid MoE offloading, new QMoE kernels and faster session init.

Release v1.32.0 · microsoft/onnxruntime
MicrosoftOpenCodeFree
GitHub · 16h #

DeepSpeed v0.20.0 released with training optimizations

DeepSpeed is Microsoft's open-source deep learning optimization library that makes distributed training and inference efficient. Version 0.20.0 adds changes including GPTBigCode shared key/value gradient summing and AutoEP expert storage maps.

Release v0.20.0 · deepspeedai/DeepSpeed
DeepSpeedOpenCodeFree
GitHub · 1d #

Hugging Face releases diffusers v0.43.0

Diffusers is Hugging Face's open-source PyTorch library for running and training diffusion models for image, video and audio generation. v0.43.0 adds five modular model integrations across image generation, editing, video with audio, speech-driven animation and world simulation, plus attention and LoRA improvements.

Release v0.43.0 · huggingface/diffusers
Hugging FaceOpenCodeOpen weights
GitHub · 1d #

FrigadeHQ releases Yogurt, open-source private AI meeting notes

Yogurt is a free, open-source TypeScript app that takes private AI meeting notes, keeping transcripts and summaries local rather than sending them to a hosted service. It drew 133 GitHub stars in its first week.

New repo · ★ 133 in its first week · TypeScript
FrigadeHQOpenCodeFree
GitHub · 1d #

LASIK CLI and MCP server undo BLA agent blinding

LASIK is a static CLI and MCP server that reverses Blind LLM Agents hardening so coding agents can read real code again, without executing the target. It reached 100 GitHub stars in its first week.

New repo · ★ 100 in its first week · TypeScript
pktqOpenCodeFreeCybersecurity
More On GitHub →

Apps for Everyone → rss

Pi · 8h #

Pi app updates to version 26.1109.0 for iPhone

Pi, a personal AI assistant app for iPhone and iPad, released version 26.1109.0 with bug fixes and performance improvements.

Version 26.1109.0 · iPhone / iPad
InflectionClosedAppFree
Chatbox · 22h #

Chatbox releases desktop app version 1.24.4

Chatbox 1.24.4 is out for macOS, Windows and Linux. It is a desktop app for chatting with GPT-4 and GPT-3.5 through the OpenAI API.

Version 1.24.4 · macOS / Windows / Linux
ChatboxClosedFreeAPI
More Apps for Everyone →Latest app versions & downloads →

Developer Tools → rss

Product Hunt · 7h #

Burnstate launches Mac menu bar app for AI coding limits

Burnstate is a native Mac menu bar app that shows Claude Code, Codex, Cursor and Vercel AI Gateway limits, estimates when you'll run out, answers Claude Code permission prompts, and tracks session tokens and cost.

Launch · ▲ 80 on Product Hunt
BurnstateAppPaid
Conductor · 16h #

Conductor 0.94.2 adds Claude Code parallelization

Conductor, a desktop app for running multiple AI coding agents in parallel, released version 0.94.2 with Claude Code parallelization on macOS, Windows and Linux.

Version 0.94.2 · macOS / Windows / Linux
ConductorClosedFree
Amazon Q · 16h #

Amazon Q 2.10.0 released for VS Code

Amazon Q, AWS's generative AI coding assistant for VS Code, updated to version 2.10.0 with unspecified fixes and improvements.

Version 2.10.0 · VS Code extension
AmazonClosedFreePaid
More Developer Tools →Latest dev tool versions & downloads →