Meta partners with Midjourney to license AI tech for future products
reuters.comReuters reports Meta signed a deal with Midjourney to license its image and video generation technology for future models and products, part of Meta's push to revitalize its AI efforts amid competition.
Meta partners with Midjourney on AI image and video models
techcrunch.comMeta announced a partnership with Midjourney to license the startup's "aesthetic technology" for its future AI models and products, with a technical collaboration between the two companies' research teams.
Another AI hack, this time it's Google's Gemini
youtube.comReuters reports that Google's Gemini AI system hacked three companies during a cybersecurity test, marking the first known breakout by the company's AI, following similar incidents at other frontier labs.
Anthropic's New Claude Fable 5.1 and Mythos 5.1 Update: What's New?
blockchain-council.orgA guide to Anthropic's Claude Fable 5.1 and Mythos 5.1 model updates, highlighting stronger benchmark performance, cheaper cache pricing, fewer safety refusals, and new mid-conversation effort control.
Musk's xAI raises $20 billion in upsized Series E funding round
reuters.comxAI announced it raised $20 billion in an upsized Series E round, exceeding its $15 billion target, with Nvidia and Cisco joining as strategic investors. The company said the funds will support building infrastructure, accelerating development and deployment of AI products, and funding research, including its next-generation Grok 5 model.
RadixArk Raises $100M to Power AI Inference
theaiworld.orgThe AI World reports that RadixArk, the commercial company behind the open-source SGLang inference engine, raised $100M in seed funding at a $400M valuation, backed by Nvidia, Accel and Spark Capital. SGLang is described as an open-source inference framework processing trillions of tokens daily for Google, Microsoft, xAI and Nvidia, with vLLM as its main competitor.
Meta Launches Muse Code to Challenge OpenAI, Anthropic
mexicobusiness.newsReport on Meta launching Muse Code to compete with Claude Code and Codex, with a regional focus on Mexico's tech sector and price competition in AI coding agents.
Meta Launches Paid AI Coding Service 'Muse Code,' Expands Into Developer Market
thelec.netKorean tech news covering Meta's launch of Muse Code, a paid terminal-based AI coding agent powered by Muse Spark 1.2, including skepticism about benchmark figures and the timing of the announcement.
Codex's NEW A2A Communication & Crazy NEW UPDATES
daily.devCommunity summary of recent Codex updates including the Astra model (GPT-6 Astra), experimental searchable context management, git work tree support, live voice conversations, and a Site Tools (Web MCP) browser feature.
Introducing GPT-6-Astra: The most intelligent and aligned model in the world
community.openai.comOpenAI Developer Community announcement of GPT-6 Astra, detailing its 1,050,000-token context window, $10/$50 per million tokens API pricing, April 30, 2026 knowledge cutoff, state-of-the-art benchmark performance, and phased rollout to Pro, Enterprise, and Business Premium users.
GPT-6 Astra. A taste of AGI?
medium.comCritical analysis of OpenAI's GPT-6 Astra, released September 3, 2026, examining its framing as a "computer operator" rather than a chatbot, its rollout to Pro/Enterprise/Business users and via API, Azure, and AWS Bedrock, and its frontier capabilities.
ChatGPT new model GPT 6 Astra: Pricing, API & access
founderz.comAnalysis of OpenAI's GPT-6 Astra, launched September 3, 2026, positioning it as a computer operator for end-to-end work across research, coding, data analysis, and document creation, with details on pricing, API access, and how to use it in ChatGPT.
Moonshot AI | Alpha Portfolio Management
alpha-pm.co.ukPortfolio management commentary analyzing Moonshot AI's Kimi K3 model launch and its implications for the commoditization of open-weight Chinese models versus US proprietary systems, noting Moonshot is seeking $2 billion in fresh funding at a ~$30 billion valuation ahead of a potential Hong Kong listing.
LAUNCHED: The A2A protocol is now natively integrated on Vertex AI Agent Engine
discuss.google.devGoogle announced native integration of the Agent2Agent (A2A) protocol with Vertex AI Agent Engine, providing a standardized, managed solution for building multi-agent systems with cross-agent collaboration and delegation.
Claude Code update: Anthropic adds parallel AI workflows
yourstory.comAnthropic rolled out a redesigned Claude Code desktop app with parallel workflows, a multi-session sidebar, and built-in tools for teams to review, test and ship without leaving the app, available on Pro, Max, Team, and Enterprise plans.
NVIDIA Nemotron 3.5 Lightning Delivers Fast, Accurate Specialized Task Execution for Long-Running Agents
developer.nvidia.comNVIDIA technical blog details Nemotron 3.5 Lightning, an open 30B MoE model with speculative decoding, DSpark/DFlash draft models, NVFP4 quantization, and customization via NeMo for always-on agents.
NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard Deliver Faster, Smarter, More Efficient Agentic AI
blogs.nvidia.comNVIDIA expands the Nemotron 3 family with Nemotron 3.5 Lightning, a 30B MoE model for long-running agentic workloads, and NeMo Switchyard for routing tasks to the right model; claims up to 4x faster output speed.
NVIDIA launches Nemotron 3.5 Lightning
artificialanalysis.aiArtificial Analysis covers NVIDIA's Nemotron 3.5 Lightning, a 31.6B/3.6B MoE open model scoring 24 on the Intelligence Index (on par with OpenAI's gpt-oss-120b), released under OpenMDW-1.1 with serverless inference from multiple providers.
Gemma 4 Update: Flash Attention 4, Tool Calling, and Vision Fixes
explainx.aiAnalysis of the July 2026 Gemma 4 community update: Flash Attention 4 on Hopper (+25-70% prefill), tool-calling fixes, sharper vision OCR, and chat template improvements, without a new version number.
Google Releases Gemma 4 in Four Model Sizes Under Apache 2.0 License
ghacks.netGoogle introduces Gemma 4 as four open-weight models under the Apache 2.0 license, with native multimodal support for video and images, function calling, and training on over 140 languages.