Zlio
Everything in AI and robotics, on one page.
Updated 06:32 UTC · Next update 07:00 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

Mashable · 2h #

Apple debuts musical AI assistant for creators

Why it matters: Apple entering AI music tools could reshape creator workflows across its devices.

Apple introduced a new AI assistant aimed at music creators and artists, pitched as a musically-inclined tool for creative work.

AppleClosed
Pandaily · 4h · 2 sources #

Meituan releases LongCat-2.5, 1.6T multimodal model for agentic coding

Why it matters: A 1.6T MoE built for long-horizon coding agents shows Chinese labs pushing agentic models.

LongCat-2.5 is a 1.6T-parameter native multimodal model with about 48B active per token, trained with long-horizon reinforcement learning for agentic coding and optimized for W4A4 inference on domestic Chinese compute.

MeituanClosedAPI
More: TechNode
DigiTimes Asia · 4h · 2 sources #

Canada to invest up to CAD1.6B in Cohere

Why it matters: A major sovereign investment in a domestic AI champion to reduce reliance on US and China.

The Canadian government plans to invest up to CAD1.6 billion (about US$1.14 billion) in domestic AI startup Cohere, becoming a major investor in its latest funding round as Canada seeks AI independence.

Cohere
More: digitimes
MediaNama · 14h · 5 sources #

Anthropic's Claude found exploits in Chrome, Android, Windows, Linux

Why it matters: Frontier models finding real exploits raises stakes for AI-powered security and defence.

An unreleased Claude model found vulnerabilities across five targets, including turning a Linux kernel bug into remote root in 53 minutes, according to reports. Anthropic has not disclosed success rates or compute costs; alleged Mythos 5.5 leaks are unverified.

AnthropicClosedCybersecurity

Stability AI Platform launches

Why it matters: Stability AI moves from licensing weights to a hosted platform, risking the open image-model community it built.

Stability AI, the company behind Stable Diffusion image models, has launched its own platform for running and building with its AI models.

Stability AIClosedAppOfficial
Axios · 16h · 17 sources #

OpenAI plans post-midterm push, rebranding data centers as AI compute campuses

Why it matters: OpenAI's political strategy around data centers could shape AI infrastructure policy.

After Tuesday's midterms, OpenAI plans a policy and PR blitz to align with the new Washington, including rebranding its data centers as 'AI compute campuses' amid public backlash.

OpenAIDatacenters
Reddit · 23h #

FineDNA releases dataset for training DNA-focused language models

Why it matters: Open DNA training data could lower barriers to building genomics language models.

FineDNA publishes the data it used to build Carbon-X, the DNA-focused model from Qwen3-14B: 300B+ genome tokens for midtraining, about 200K SFT conversations and 98K RL prompts, with code and recipe open.

FineDNAOpenOpen weightsCodeFree
TechNode · 1h #

Apple to start China Apple Intelligence beta with Alibaba's Qwen

Why it matters: Apple Intelligence finally enters China, pairing iPhones with Alibaba's Qwen models.

Apple will begin testing Apple Intelligence in mainland China on November 4, integrating Alibaba's Qwen into Siri, Writing Tools and Visual Intelligence via an iOS developer beta, ahead of a wider iOS 27.3 release in December.

AppleClosedAppWaitlist
Reuters · 7h · 11 sources #

China's robot makers shift race from hardware to AI brains

Why it matters: The physical-AI software layer is becoming the key battleground for humanoid robots.

Reuters reports Chinese humanoid robot makers have mastered locomotion and are now competing on intelligence software and brains, with focus shifting to job-ready robots that can perform real work.

BusinessKorea · 23h Widely covered · 10 sources #

Samsung chair Lee visits Yokohama packaging lab for AI chip edge

Why it matters: Advanced packaging is a key AI chip bottleneck, and Samsung is investing heavily in Japan.

Samsung is investing ¥40B (about $253.5M) over five years in its Advanced Package Lab in Yokohama, Japan. Chairman Lee Jae-yong visited the packaging research hub to boost AI chip materials and equipment cooperation.

SamsungChips
The Information · 1d Widely covered · 53 sources #

Meta built humanoid robot prototype with China's Fourier

Why it matters: Meta entering humanoid robotics with a Chinese partner signals physical AI competition.

Meta reportedly built a 106-pound humanoid robot named Max in three months with Chinese startup Fourier, as part of a push to ramp up its robotics efforts.

Meta
Mashable · 1d #

Apple opens Siri AI beta with new things to try

Why it matters: Siri's AI overhaul is Apple's biggest assistant change and affects hundreds of millions of iPhones.

Mashable lists seven things to try in the available beta of Apple's AI-upgraded Siri, covering new capabilities users can test now.

AppleClosedFree
GitHub · GitHub · 5h #

NextTransformer open-source framework targets next-gen Transformer research

NextTransformer is an open-source Python framework for next-generation Transformer architectures, offering compute-matched benchmarking and a platform for autonomous architecture research. It drew 101 stars in its first week.

New repo · ★ 101 in its first week · Python
Interactive-Active-AIOpenCodeFree
Dev tool · Slashdot · 5h #

Stack Overflow says AI hasn't killed it

Stack Overflow's senior VP of community says reports of the site's death are exaggerated, arguing AI changed how developers access knowledge but the knowledge and company behind it remain.

Stack Overflow
Dev tool · GitHub · 7h #

FlashAlign offers per-request model alignment

FlashAlign is a Python tool that re-aligns AI models instantly, either per request for verified users and channels or always on, and works with Hermes, OpenClaw and custom harnesses. It reached 100 stars in its first week.

New repo · ★ 100 in its first week · Python
AevonixOpenCodeFree
Dev tool · GitHub · 10h #

Balanov/uncut releases 7 MB lossless coding-agent harness

uncut is a 7 MB coding-agent harness for shell, CI and local models that claims lossless context with no silent truncation and no TUI. Written in Go, it gained 121 GitHub stars in its first week.

New repo · ★ 121 in its first week · Go
BalanovOpenCodeFree
Hugging Face · Hugging Face · 14h #

Phonon browser TTS demo runs five languages locally

Phonon is an open text-to-speech model family; this Hugging Face Space demo runs it entirely in your browser in English, French, German, Spanish and Portuguese, with no audio leaving the device.

Space · static · ♥ 100
gradiumOpenOpen weightsFree
Dev tool · CodeBuddy · 15h #

CodeBuddy releases 4.12.2 adaptive AI IDE

CodeBuddy is an AI-powered adaptive IDE that adjusts its coding assistance to the developer's workflow. Version 4.12.2.41207131 is released for macOS, Windows and Linux.

Version 4.12.2.41207131 · macOS / Windows / Linux
CodeBuddyClosedFree
Dev tool · GitHub · 15h #

AnyModelBot open-sources local-first AI bots for any model

AnyModelBot is an open-source, local-first alternative to hosted AI bots that runs agents on 70+ models via one Modelflare key, reachable in-app, WeChat or Feishu, with memory and computer control.

New repo · ★ 100 in its first week · TypeScript
ModelflareOpenCodeFree
Dev tool · GitHub · 16h #

graphmemory offers bitemporal AI memory with decay and dedupe

graphmemory is an open-source memory system for AI apps: it stores facts bitemporally, applies time-graded recall and decay, expires entries by TTL and dedupes, backed by Neo4j, PostgreSQL and Redis.

New repo · ★ 102 in its first week · Python
graphmemoryOpenCodeFree
Dev tool · GitHub · 18h #

Tucano GitHub project launches local multi-agent coding teams

Tucano is a GitHub project for running a team of AI agents on your Mac, Windows or Linux machine. Agents talk in channels, write and review code, and use your tools on Claude Code, Codex, OpenCode or Pi; risky actions wait for approval.

New repo · ★ 100 in its first week · TypeScript
TucanoOpenCodeFree
All app versions & downloads → All dev tool versions →
Older news → RSS →

Shipping now: what you can use today

On Hugging Face → rss

Hugging Face · 14h #

Phonon browser TTS demo runs five languages locally

Phonon is an open text-to-speech model family; this Hugging Face Space demo runs it entirely in your browser in English, French, German, Spanish and Portuguese, with no audio leaving the device.

Space · static · ♥ 100
gradiumOpenOpen weightsFree
Hugging Face · 1d #

Hugging Face demo offers realtime Qwen-Image-2.1 editing

A Hugging Face Space demos realtime text-to-image and multi-image editing with Qwen-Image-2.1, using a 2-step DMD2-distilled LoRA fused into the weights and AOTInductor kernels on ZeroGPU, streaming 512px results live.

Space · gradio · ♥ 100
QwenOpenFreeCode
More On Hugging Face →

On GitHub → rss

GitHub · 7h #

FlashAlign offers per-request model alignment

FlashAlign is a Python tool that re-aligns AI models instantly, either per request for verified users and channels or always on, and works with Hermes, OpenClaw and custom harnesses. It reached 100 stars in its first week.

New repo · ★ 100 in its first week · Python
AevonixOpenCodeFree
GitHub · 10h #

Balanov/uncut releases 7 MB lossless coding-agent harness

uncut is a 7 MB coding-agent harness for shell, CI and local models that claims lossless context with no silent truncation and no TUI. Written in Go, it gained 121 GitHub stars in its first week.

New repo · ★ 121 in its first week · Go
BalanovOpenCodeFree
GitHub · 15h #

AnyModelBot open-sources local-first AI bots for any model

AnyModelBot is an open-source, local-first alternative to hosted AI bots that runs agents on 70+ models via one Modelflare key, reachable in-app, WeChat or Feishu, with memory and computer control.

New repo · ★ 100 in its first week · TypeScript
ModelflareOpenCodeFree
GitHub · 16h #

graphmemory offers bitemporal AI memory with decay and dedupe

graphmemory is an open-source memory system for AI apps: it stores facts bitemporally, applies time-graded recall and decay, expires entries by TTL and dedupes, backed by Neo4j, PostgreSQL and Redis.

New repo · ★ 102 in its first week · Python
graphmemoryOpenCodeFree
GitHub · 18h #

Tucano GitHub project launches local multi-agent coding teams

Tucano is a GitHub project for running a team of AI agents on your Mac, Windows or Linux machine. Agents talk in channels, write and review code, and use your tools on Claude Code, Codex, OpenCode or Pi; risky actions wait for approval.

New repo · ★ 100 in its first week · TypeScript
TucanoOpenCodeFree
GitHub · 1d #

NamiAgent tutorial repo teaches building a local AI agent

A zero-to-one beginner tutorial that doubles as a runnable local agent: file and command access, MCP, long-term memory and a web UI showing each step. Builds up from a 100-line loop and dissects Claude Code and Codex.

New repo · ★ 101 in its first week · Python
NamiAgentOpenCode
GitHub · 1d #

Chinese-language Stanford CS329Z AI agents course posted to GitHub

A Chinese translation of Stanford's CS329Z Engineering AI Agents course for Python developers, with 19 lectures and four practicals covering LLMs, RAG, tool use, memory, multi-agent systems, evaluation and safety, in Markdown and HTML.

New repo · ★ 103 in its first week · HTML
Stanford CS329ZOpenFreeCode
More On GitHub →

Apps for Everyone → rss

More Apps for Everyone →Latest app versions & downloads →

Developer Tools → rss

Slashdot · 5h #

Stack Overflow says AI hasn't killed it

Stack Overflow's senior VP of community says reports of the site's death are exaggerated, arguing AI changed how developers access knowledge but the knowledge and company behind it remain.

Stack Overflow
CodeBuddy · 15h #

CodeBuddy releases 4.12.2 adaptive AI IDE

CodeBuddy is an AI-powered adaptive IDE that adjusts its coding assistance to the developer's workflow. Version 4.12.2.41207131 is released for macOS, Windows and Linux.

Version 4.12.2.41207131 · macOS / Windows / Linux
CodeBuddyClosedFree
Reddit · 1d #

Ultracode now pairs with any effort level

A Reddit PSA says Ultracode is no longer locked to max effort: users can pick Low effort with Ultracode on, keeping multi-agent workflow orchestration while cutting thinking cost.

Paid
Simon Willison · 1d #

Datasette-llm 0.1a9 adopts LLM's default model

Datasette-llm is a plugin that adds LLM-powered features to Datasette, a tool for exploring and publishing data. Version 0.1a9 now picks up the default model used by LLM, letting users start Datasette Agent without extra configuration.

DatasetteOpenCodeFree
More Developer Tools →Latest dev tool versions & downloads →