Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Everything Is a Stream: runtime composability over compile-time plugins(antigma.ai ↗)
    discuss
  2. JRR Tolkien on 'the most tragic moment' in The Lord of the Rings(bbc.com ↗)
    discuss
  3. Jev is good for agentic security use cases too(vega.io ↗)
    discuss
  4. Confectionery: A lightweight build configuration system(github.com/a5x0 ↗)
    1comments
  5. Paramount to Resolve Antitrust Lawsuit, Clearing Way for Warner Bros Acquisition(nytimes.com ↗)
    discuss
  6. SAML: A Fractal of Bad Design(trailofbits.com ↗)
    discuss
  7. Show HN: AlarmK – a mission alarm for iPhone, built on iOS 26's AlarmKit(apps.apple.com ↗)
    discuss
  8. Show HN: Jev AI –Try the Interactive Playground and Typed Decision(jev-ai.info ↗)
    discuss
  9. Thoughts on Techno-Anthropology(medium.com/kazarazat ↗)
    discuss
  10. Running an Optimal Trace(anishathalye.com ↗)
    discuss
  11. Show HN: FractalBrainOS – self-learning brain in C++17 (no backprop)(github.com/nineni999nenine ↗)
    discuss
  12. White House press pool (5 networks) suspends coverage(nytimes.com ↗)
    1comments
  13. Show HN: Mathy – Build Math Automaticity(mathy.game ↗)
    discuss
  14. Skill and visualisation kit to help developers understand PRs better(github.com/annanay25 ↗)
    1comments
  15. OpenClaw Completes Security Audit(openclaw.ai ↗)
    discuss
  16. U.S. Site Blocking Bill Adds VPNs to the List of Blocking Intermediaries(torrentfreak.com ↗)
    discuss
  17. Show HN: HN for Me – Jev curates Hacker News based on your interests(github.com/raahelpie ↗)
    discuss
  18. Practical model evaluation and compression tools(github.com/0xsero ↗)
    discuss
  19. FUSE over Io_uring(camandro.org ↗)
    discuss
  20. Aikido Altar: open-weight security model(aikido.dev ↗)
    1comments
  21. A study of sequence weighting at scale(janestreet.com ↗)
    discuss
  22. The Biopolitics of Intelligence(muskdeer.blogspot.com ↗)
    discuss
  23. Stop using your most expensive model for every decision(twitter.com/usebuddy ↗)
    discuss
  24. Show HN: PokerTools Arena – Local AI vs. AI Poker LLM Benchmark Table(github.com/pokertools-arena ↗)
    discuss
  25. One Canvas Many Hands(onecanvasmanyhands.com ↗)
    discuss
  26. Drone WMDs Don't Need Any New Technology(lesswrong.com ↗)
    discuss
  27. Scientists heard a knock at the door. It was a pelican in need of help.(washingtonpost.com ↗)
    discuss
  28. Using Jev to substitute inverse kinematics(twitter.com/caloriepaper ↗)
    discuss
  29. What turned out to be hard about the VulpieAI overlay(medium.com/vulpie.ai.support ↗)
    discuss
  30. Trust Fall: a small browser game about spending trust(ilands.ai ↗)
    discuss

Step 5 Preview: StepFun's frontier model with 1M context and video input

2 pointsby 1h agoplatform.stepfun.ai
1 comments
1h agoHN ↗

Anyone tried this for actual coding yet? Curious how token-effective it is compared to Kimi K3. The benchmark results looks close but in real-world tasks it may different.