Moonshot AI releases new Kimi model; concerns raised about 'full AI communism
Chinese company Moonshot AI released a new version of its Kimi model this week, prompting concern about "full AI communism."
NVIDIA Nemotron 3 Ultra achieves benchmark-leading performance at lower cost than top closed models; LangChain's Deep Agents harness achieves highest accuracy among open models with higher throughput and more tasks.
NVIDIA Nemotron 3 Ultra is offering leading performance at lower cost than top closed models with the largest and most widely adopted AI agent orchestration platform. LangChain tuned its Deep Agents harness for...
X.ai makes Grok Build open source; introduces Grok 4.5
Grok Build is Now Open Source X.AI
OpenAI builds GPT-Red, an LLM sparring partner to improve model defenses; releases GPT-5.6, claiming robustness improvement
OpenAI has built an LLM super-hacker called GPT-Red that it uses as a sparring partner to help its other models boost their defenses against cyberattacks. Last week the company released the latest version of its...
Thinking Machines Lab releases Inkling, a 975-billion-parameter open source model trained on video and audio.
Inkling, a 975-billion-parameter open source model, was trained to understand video and audio. It could help Thinking Machines establish itself among competitors like Anthropic and OpenAI.
Moonshot AI launches Kimi K3 with 2.8 trillion parameters, calling it the first open 3T-class model; DeepSeek previously held the title.
Chinese AI lab Moonshot AI announced Kimi K3 this morning, describing it as their "most capable model to date, with 2.8 trillion parameters". It's currently available via their website and API, but an open weight...
Nvidia reduces GPU memory bottlenecks in JAX-based LLM training via host offloading.
Large language model (LLM) training workloads increasingly run into GPU memory limits before compute is fully used. Model weights, gradients, optimizer states,...
Inkling is a Mixture-of-Experts transformer with 975B total parameters and open-weights; Mira Murati's Thinking Machines Lab releases the model and plans Inkling-Small with 276B parameters.
Inkling: Our open-weights model Mira Murati's Thinking Machines Lab just released their first open-weights model. Inkling is "a Mixture-of-Experts transformer with 975B total parameters, 41B active" - an Apache-2.0...
Apple files trade secrets lawsuit against OpenAI, alleging misconduct and former employees; OpenAI's response is hedged as IPO timing is problematic.
Apple filed a trade secrets lawsuit against OpenAI last Friday, and it’s not messing around. The complaint alleges a pattern of misconduct reaching all the way up to OpenAI’s chief hardware officer and claims more...
EU to force Google to share search data and open AI on Android
Google says these changes could endanger user privacy and security.
OpenAI releases first hardware; ChatGPT basketball mentioned.
You may have heard that OpenAI released its first piece of hardware this week. You may not have heard about the ChatGPT basketball.
Moonshot's Kimi K3 is the largest open AI model from China with 2T-3T parameters; it's expected to close the gap with Anthropic's Opus 4.8
The FT reports Kimi K3 will be the largest open AI model from China, with a parameter count between 2 trillion and 3 trillion.
NVIDIA's Vera Rubin achieves low cost per token via extreme codesign, maximizing intelligence per dollar for post-training workloads in agentic AI.
Lowest cost per token from extreme codesign maximizes intelligence per dollar for post-training in the agentic era.
Grok CLI codebase includes Mermaid to Unicode box art tool; newly open-sourced
Tool: Mermaid to Unicode box art (grok-mermaid) While exploring the codebase for the newly open-sourced Grok CLI coding agent I came across xai-grok-markdown/src/mermaid.rs, a "self-contained terminal...
Google DeepMind and Isomorphic Labs share joint approach to bioresilience and AI models.
Google DeepMind and Isomorphic Labs are sharing our joint approach to bioresilience and AI models.
Microsoft is reportedly training salespeople to position its in-house AI models as more efficient and cost-effective than OpenAI and Anthropic's.
Microsoft is looking to sell its in-house AI models as more efficient and cost-effective than its competitors' models.
Period tracker likely collects user data; Russian cyberattackers shift to infrastructure hacking; DHS repeatedly fails to detect breaches; AI music generator breach reveals scraping practices.
Plus: Russian cyberspies turn to infrastructure hacking, DHS repeatedly fails to realize it’d been hacked, a breach exposes an AI music generator’s scraping ways, and more.
Prompt injection attacks using context bombing trick malicious AI agents into shutting down.
“Context bombing” tricks malicious AI agents into shutting down before they can do harm.
Neil Rimer predicts redistribution of AI-generated wealth in Silicon Valley
Neil Rimer, the venture capitalist who co-founded Index Ventures, predicts the historic wealth AI is generating in Silicon Valley will have to be redistributed, voluntarily or involuntarily.
Disney Jr. and AI animation outfit Animaj release 'Ozzy Fox' on YouTube.
Created by Jennifer Oxley and developed by 'Pocoyo' creator Guillermo García-Carsí, the preschool series has drawn nearly 800,000 views in its first two days.
NVIDIA Vera CPU Enhances AI Factory Throughput for Agentic Workloads
Agentic systems turn model reasoning into action through multi-step workflows that combine inference, tool use, code execution, retrieval, orchestration, and...
OpenAI launches GPT-5.6 family with Luna, Terra, Sol; pricing details provided.
OpenAI's latest flagship model hit general availability this morning, and comes in three sizes: Luna, Terra, and Sol (from smallest to largest). The new models are priced per 1M input/output tokens as Luna $1/$6,...
Google Vids adds Gemini Omni and Personal Avatars for video creation and editing
Gemini Omni and personal avatars in Google Vids make video creation easier than ever.
OpenAI introduces GPT-Red, an automated red teaming system using self-play to enhance AI safety, alignment, and prompt injection robustness.
Explore GPT-Red, OpenAI’s automated red teaming system that uses self-play to improve AI safety, alignment, and prompt injection robustness.
NVIDIA Nemotron Model Reasoning Challenge; 5,000+ Kagglers explored techniques to improve AI reasoning accuracy.
The NVIDIA Nemotron Model Reasoning Challenge invited the Kaggle community to explore a focused question: What techniques can improve reasoning accuracy when...
Microsoft 365 Copilot now uses GPT-5.6 as preferred model; it powers Word, Excel, PowerPoint, Chat, and Cowork with stronger AI capabilities.
Learn how GPT-5.6 powers Microsoft 365 Copilot with stronger AI capabilities across Word, Excel, PowerPoint, Chat, and Cowork for faster, higher-quality work.
Stray Sheep uses ElevenLabs to enhance accessibility for Deaf people in Bulgaria.
How Stray Sheep uses ElevenLabs to make Bulgaria more accessible for Deaf people ElevenLabs
Anthropic extends Fable 5 access and increases Claude Code rate limits through July 19.
One of the consequences of GPT-5.6 Sol being clearly a Fable/Mythos class model is that Anthropic have, once again, bumped the date that Fable stops being available in their Claude Max plans: We're extending Claude...
Linus Torvalds responds to AI coding critics in Linux by urging them to fork the code or walk away.
Creator says he will "very loudly ignore" those arguing for a ban on AI tools.
Roblox launches AI-powered "Build" feature in mobile app for text prompt game creation.
Roblox's new "Build" feature lets users generate basic games using a single text prompt.
OpenAI releases $230 light-up keyboard for Codex; legal battle with Apple over hardware trade theft ongoing
OpenAI, which is in the middle of a legal battle with Apple over hardware trade theft allegations, just released a light-up keyboard designed to be paired with its agentic coding app.
Google updates Gemini usage tracking; fewer AI responses possible
Now that Google has changed how its usage quotas are tallied, you might not get as many AI responses as you did before.
Gemini builds Managed Agents using GitHub CLI without exposing PAT; egress proxy injects real token while sandbox uses dummy.
Build Gemini Managed Agents that use the GitHub CLI without exposing your PAT. The egress proxy injects your real token into outbound requests while the sandbox only sees a dummy token.
xAI can’t deny Grok makes CSAM anymore. So it’s suing users.
Elon Musk's xAI files first lawsuit against Grok user accused of making child sex images.
OpenAI enhances ChatGPT safety for teens
Learn how OpenAI is making ChatGPT safer for teens with age-appropriate protections, learning tools, parental controls, and expert partnerships.
llm 0.31.1 fixes bug in OpenAI Chat Completion endpoints with empty tool arguments; tested in llm-meta-ai
Release: llm 0.31.1 Fix for a bug with OpenAI Chat Completion endpoints where a tool call with empty arguments could result in a JSON error from some providers. #1521 This bug came up when I was testing...
Former DeepMind researcher Andrew Dai raises $300M pre-seed for visual AI product; prior work influenced ChatGPT development.
Drawing on more than a decade spent helping build some of the world's most influential AI systems, including research that later informed the development of ChatGPT, Andrew Dai explains why he believes visual AI is...
Tool converts Mermaid to ASCII art using Grok Build's Rust code; older Go library AlexanderGrooff/mermaid-ascii exists
Tool: Mermaid to ASCII art (mermaid-ascii) After building the Mermaid to ASCII tool based on Grok Build's Rust code I learned that there's an older, more fully-featured Go library called...
Google and AIM launched ATL Saathi, a Gemini-powered AI tool for Indian educators in robotics labs.
Google and AIM launched ATL Saathi, a Gemini-powered AI tool empowering Indian educators in robotics labs.
OpenAI's Bio Bounty program targets GPT-5.5
Details about the OpenAI Bio Bounty program
Google-backed satellites for wildfire detection launch as smoke chokes US, Canada; FireSat can spot wildfires that other satellites miss
The FireSat program can spot wildfires that other satellites miss.
AI Model Co-Design Prioritizes Hardware-Optimized Performance Metrics
AI performance comes down to three dimensions: Accuracy: How well the model reasons and produces outputs Throughput: How many tokens per second a...
Google adds AI avatars to Vids for personalized video creation; Gemini Omni tools enable prompt-based video generation and editing.
Google is adding personalized AI avatars to Vids that let users create videos starring a digital version of themselves, alongside Gemini Omni-powered tools for generating and editing videos from prompts and reference...
Alexandre LeBrun of AMI Labs rejects labeling AI as AGI or superintelligence; industry pursues superintelligence goals
While everyone in AI is chasing "superintelligence," Alexandre LeBrun, CEO of Yann LeCun’s world model startup, AMI Labs, dismisses the word.
Anthropic develops a Jacobian lens to probe inside Claude and reveals what large language models do when answering questions and performing tasks
The AI firm Anthropic has developed a technique that has given it the clearest glimpse yet at what’s really going on inside large language models as they answer questions or carry out tasks. What they found ranges...
Agility Robotics opens new training center for Digit robots in Fremont, California.
Agility is opening a new training center for its Digit robots in Fremont, California.
Sarah Friar, OpenAI CFO, introduces AI scorecard to measure ROI via useful work, cost per task, dependability, and compute return.
Sarah Friar, CFO of OpenAI, introduces a practical AI scorecard to measure ROI through useful work, cost per successful task, dependability, and return on compute.
Apple sues OpenAI over legal and reputational issues
On today’s Uncanny Valley, we unpack OpenAI’s ongoing drama, both legal and reputational, and whether these developments could further hurt the company—particularly in its fight against Anthropic.