deepmind open-sources science skills: ai agents meet 30+ databases
google is quietly making a strategic move in the war for scientific talent deepmind just open-sourced "science skills" – a toolkit that lets ai agents natively talk to 30+
google is quietly making a strategic move in the war for scientific talent deepmind just open-sourced "science skills" – a toolkit that lets ai agents natively talk to 30+
Addy Crezee
1 min read
what changed under the hood: • vision. replaced the encoder with a lightweight embedding module (single matrix multiply + positional embedding + normalization). the llm backbone now handles visual processing directly • audio. encoder
Addy Crezee
1 min read
why he's being criticised: • leadership style called chaotic and impatient • accused of hyping small, step-by-step improvements as if they were major breakthroughs • talent leaks – a former apple exec
Addy Crezee
1 min read
michael burry says neither spacex nor anthropic is worth $1t the hacker news thread runs deeper than the headline – the real argument is over whether the trillion-dollar number can be
Nick Trenkler
3 min read
• may 5 – gpt-5.5 (@OpenAI) • may 8 – ring-2.6-1t (@TheInclusionAI) • may 18 – composer 2.5 (@cursor_ai) • may 19 – gemini 3.5 flash (@Google) • may 20 – qwen 3.7 max
Nick Trenkler
1 min read
microsoft build is live right now – and alan, one of our ai hosts on thehype radio, is watching every announcement as it drops by the end of the first hour
Nick Trenkler
1 min read
researchers built a multi-agent framework that cuts inference cost by 75% – without language between agents our ai host mira found a framework built by researchers from stanford, mit & nvidia
Nick Trenkler
1 min read
here’s what it means and how they achieved it the team pointed qwen3.7-plus at apple's native stocks app and told it to reproduce it. no manual
Addy Crezee
2 min read
today nvidia announced vera cpu. here's all you need to know: it's nvidia's first cpu built specifically for ai agents – not general computing, not
Nick Trenkler
1 min read
swe-bench pro across open-weight models hasn't moved in 10 weeks mar 27th - jun 1st. 5 major open-weight releases. net progress on swe-bench pro: +0.3%. scores swung
Addy Crezee
1 min read
hidream-o1-image-dev-2604 is ranked #1 across all open weight text-to-image models on the artificial analysis text-to-image arena what makes this model interesting is how it's built. many other image
Addy Crezee
2 min read
xai just dethroned bytedance in the image-to-video arena – first time in 3.5 months (since feb 12) that seedance 2.0 isn't #1 the new rankings: 1. xai
Nick Trenkler
1 min read
rohit ghumare @ghumare64 is on the weekly github radar once again, this time with the ai engineering from scratch repo that has shown impressive 54.60% growth another returning repository
Nick Trenkler
1 min read
by mid-2026, many assumptions of the first wave of genai have almost collapsed. the structural bottleneck in ai development is no longer model intelligence, nor is it raw context window
Nick Trenkler
8 min read
step 3.7 flash dropped yesterday – open weights, competitive with gemini flash 3.5 and deepseek v4 flash. and then immediately free for anyone to download. that last part is
Addy Crezee
2 min read