Meta to add 100MW of solar power from US gear
The social media company is adding another tranche of solar to power a new AI
Stop benchmarking in the lab: Inclusion Arena shows how LLMs perform in production
Researchers from Inclusion AI and Ant Group proposed a new LLM leaderboard that takes its
LLMs generate ‘fluent nonsense’ when reasoning outside their training zone
Chain-of-Thought isn’t a plug-and-play solution. For developers, this research offers a blueprint for LLM testing
DeepSeek V3.1 just dropped — and it might be the most powerful open AI yet
China’s DeepSeek has released a 685-billion parameter open-source AI model, DeepSeek V3.1, challenging OpenAI and
Qwen-Image Edit gives Photoshop a run for its money with AI-powered text-to-image edits that work in seconds
Qwen-Image-Edit caters to professionals who need control while remaining approachable for casual experimentation.Read More
Keychain raises $30M and launches AI operating system for CPG manufacturers
Keychain states it’s currently being used by top CPG brands and food retailers including 7-Eleven,
Nvidia releases a new small, open model Nemotron-Nano-9B-v2 with toggle on/off reasoning
Developers are free to create and distribute derivative models. Importantly, Nvidia does not claim ownership
Hugging Face: 5 ways enterprises can slash AI costs without sacrificing performance
Ultimately, model makers and enterprises are focusing on the wrong issue: They should be computing
GEPA optimizes LLMs without costly reinforcement learning
Moving beyond the slow, costly trial-and-error of RL, GEPA teaches AI systems to learn and
TensorZero nabs $7.3M seed to solve the messy world of enterprise LLM development
TensorZero raises $7.3 million to build an open-source AI infrastructure stack that helps enterprises scale
The looming crisis of AI speed without guardrails
The future will arrive with or without our guardrails. We must design AI’s structures now
Teaching the model: Designing LLM feedback loops that get smarter over time
How to close the loop between user behavior and LLM performance, and why human-in-the-loop systems
Vibe Coding is Shoot-and-Forget Coding
TL;DR Vibe coding is great for quick hacks; lasting software still needs real engineers. Vibe
This researcher turned OpenAI’s open weights model gpt-oss-20b into a non-reasoning ‘base’ model with less alignment, more freedom
Morris found it could also reproduce verbatim passages from copyrighted works, including three out of
The Download: Taiwan’s silicon shield, and ChatGPT’s personality misstep
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s