frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

KradleBench: Multi-agent AI evals in Minecraft

https://www.kradle.ai
1•jamest•8m ago
Hi HN -

This is James and Kemal. In past lives we led Firebase and Tensorflow/JAX respectively. HN has been a particularly big part of my (James’) career and the trajectory I’ve taken would be meaningfully different without it. Thank you!

Our team shipped a benchmark: KradleBench. We test AI in Minecraft Worlds.

1. The worlds are multi-agent. Humans and other AIs are in there too, often pursuing their own goals. Collaboration and conflict skills fall out of this property.

2. The world are dynamic. They constantly change, which means that agents need to react. “Capture the Flag” has a single winning condition, and many different ways to achieve it. This leads to many emergent and creative strategies.

3. The worlds are real-time. They are not turn-based, they do not wait for AIs to finish thinking. Nature balanced speed and intelligence in humans, we can understand this in AIs too.

Underneath all this is Kradle, a platform to test and understand intelligence. It’s a petri dish where any intelligence type (LLM, World Model, Human) can be compared and generate useful information to improve current systems, whether it’s on capabilities or safety.

Some findings so far:

- A skilled human was #1, until yesterday when Astra 6.2 surpassed them. We lose badly on solo tasks and out-compete on multi-agent ones

- Astra 6.2 iterates on its strategy 4x faster than Fable 5.2. The Claude family tends to think hard and stick with strategies. The OpenAI models tend to be more reactive and iterative, which serves them well in multi-agent, dynamic, real-time evals.

We’ve got videos on all the top model’s page, and we’ll be producing them when a new model launches. Sign up on the site to get them and for when the platform opens.

Thanks for reading,

James & Kemal

Just say no to Anthropic's IPO

https://garymarcus.substack.com/p/just-say-no-to-anthropics-ipo
1•Betelbuddy•58s ago•0 comments

Inverted-Mode Scanning Tunneling Microscopy for Atomically Precise Fabrication

https://www.somewhereville.com/2026/11/17/inverted-mode-scanning-tunneling-microscopy-for-atomica...
1•gene-h•1m ago•0 comments

MapQuest's Contrarian Comeback

https://www.theguardian.com/us-news/2026/nov/18/mapquest-google-apple-navigation-app-flock
1•twilightzone•1m ago•0 comments

No click required: How the ad industry became a pipeline for mercenary spyware

https://www.csoonline.com/article/4236143/no-click-required-how-the-ad-industry-became-a-pipeline...
1•speckx•1m ago•0 comments

Weight-Space Programming builds neural networks like a 1990s researcher

https://notesbylex.com/weight-space-programming-1990s.html
1•gmays•2m ago•0 comments

CVE Statistics 2026: 86,533 Records So Far and What Is Driving Them

https://brandefense.io/blog/cve-statistics-2026-why-vulnerability-volume-is-surging/
1•o_-_o•3m ago•0 comments

Ask HN: Hacking on the Open Metaverse

1•jvanveen•3m ago•0 comments

Germany, Austria demand TikTok's European platform be sold to owners in EU

https://www.politico.eu/article/germany-and-austria-launch-bid-for-eu-tiktok-takeover/
1•vinni2•5m ago•0 comments

Tesla's $30k Cybercab comes with a $299 monthly catch

https://www.wired.com/story/tesla-cybercab-is-the-subscription-hell-they-told-you-not-to-worry-ab...
1•DamnInteresting•6m ago•1 comments

KradleBench: Multi-agent AI evals in Minecraft

https://www.kradle.ai
1•jamest•8m ago•0 comments

Advancing chip design with OpenAI models

https://openai.com/index/advancing-chip-design-with-our-models/
2•vighneshiyer•9m ago•0 comments

Workers Are Not Going to Be OK

https://www.nytimes.com/2026/11/17/opinion/ai-bubble-workers.html
2•reasonableklout•9m ago•1 comments

String Art Font (2017)

https://erikdemaine.org/fonts/stringart/
1•akkartik•12m ago•0 comments

The tool Russia hasn't been able to replicate

https://novayagazeta.eu/en/articles/2026/11/18/the-tool-russia-hasnt-replicated-en
1•giuliomagnifico•12m ago•0 comments

Show HN: Goldcrest, a household financial model you fill in by talking

https://app.getgoldcrest.com
1•tralfamadorian8•12m ago•0 comments

What is the shape of a paid skill? – Skillfully

https://skillsthatsellsh.substack.com/p/what-is-the-shape-of-a-paid-skill
2•gerroo•12m ago•0 comments

LarpLess

https://larpless.app/
3•ggigelmo•13m ago•0 comments

Microsoft Outlook down? So was our homepage

https://bengalaviz.com/microsoft-outlook-down-so-was-our-homepage/
2•goldfingeroo7•13m ago•0 comments

Forking TuxGuitar into a 17-EDO tab editor

https://frogwizard.bearblog.dev/tuxguitar-17edo/
2•speckx•13m ago•0 comments

Lakegres: Vector search on existing Open/Hudi/Iceberg tables

https://www.onehouse.ai/blog/lakegres-vector-search-on-open-table-formats
7•cloud8bits•13m ago•2 comments

Confronting the Limits of Recursive Self-Improvement

https://www.techpolicy.press/confronting-the-limits-of-recursive-self-improvement/
2•JustinHendrix•13m ago•0 comments

GraalVM fixed the java4shell bug in September

https://github.com/oracle/graal/commit/13a4853144b802b8c5943754d179c67229f50fff
2•knuckleheads•14m ago•0 comments

Ask HN: What's your motivation to create a content website in the times of AI?

1•Gooblebrai•14m ago•0 comments

The Paramount-Warner Brothers Deal Could Usher in Another Age of Lost Media

https://www.eff.org/deeplinks/2026/11/paramount-warner-brothers-deal-could-usher-another-age-lost...
1•hn_acker•15m ago•0 comments

EFF's AI Cybersecurity Policy Framework

https://www.eff.org/deeplinks/2026/11/effs-ai-cybersecurity-policy-framework
2•hn_acker•16m ago•0 comments

Show HN: Spooky Particles, sends failed Vercel deploys to your coding agent

https://spookyparticles.com
2•n8peace•17m ago•0 comments

GTA6 is the highest rated game of all time on Metacritic

https://screenrant.com/gta-6-metacritic-score-highest-ever/
2•illidan1000•17m ago•0 comments

Alfred 6 Is Here

https://www.alfredapp.com/blog/announcements/alfred-6-is-here/
2•jiehong•20m ago•0 comments

Standard Kernel Used AI to Make Intel's Abandoned Gaudi2 Outperform Nvidia H100

https://standardkernel.com/blog/intel-gaudi2-outperforms-h100/
3•anneouyang•20m ago•0 comments

Lumi, from learning RAG to an open-source, self-hosted chatbot for agencies

https://github.com/nieve-consulting/lumi
1•arroyo•21m ago•0 comments