Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. How to run your own prediction market to bet with friends(vancouverpx.ca)
    discuss
  2. "Slop Grenades"(twitter.com/shaneparrish)
    discuss
  3. Project OM10 – open-source Mechanical Watch(openmovement.org)
    discuss
  4. The Phase-Change in User-Faced Computing(liuliu.me)
    discuss
  5. How Meta's Muse works, revealed by the 6.8 GB filesystem it sent me(mouse.dev)
    1comments
  6. Show HN: I added MCP support to Founder.best for agentic product discovery(founder.best)
    discuss
  7. Grok 4.7 from xAI, added to Personality Bench(earthpilot.ai)
    1comments
  8. Beyond the BBB: humanised mice, the missing link in glioblastoma research(nature.com)
    discuss
  9. Building standards for the next phase of AI(openai.com)
    discuss
  10. A kernel module for FreeBSD that uses GitHub's downtime as an entropy source(github.com/dominotree)
    discuss
  11. Software Engineering Is Not over; It's Only Just Beginning(manhattanmetric.com)
    discuss
  12. What If We Put Jev AI in the State Machine?(seifi.org)
    discuss
  13. A Cheap and Easy Control Screen for Home Assistant(hackaday.com)
    1comments
  14. Newborn girls and boys show different attention patterns within hours of birth(medicalxpress.com)
    discuss
  15. Pytrace: J-Trace Python SDK for hardware tracing(github.com/embedder-dev)
    discuss
  16. Designing Silicon from Scratch [video](youtube.com)
    discuss
  17. Aph – Workspaces first web browser built on stock Firefox(github.com/aph-browser)
    discuss
  18. Show HN: OpenMCP – An open, code-first fork of the Model Context Protocol(github.com/enclawed)
    discuss
  19. The Spy, the Human Computer, and the H-bomb (2013)(nuclearsecrecy.com)
    discuss
  20. Forecasting Household Power Use(stochastic.blog)
    discuss
  21. Show HN: Brig – A MicroVM sandbox for AI coding agents on Mac and Linux
    discuss
  22. Can Jev be jailbroken?(tanh.xyz)
    discuss
  23. MiMo-v2.6 is "simply" the best(twitter.com/rasbt)
    discuss
  24. The Interfaces Are Arriving(oreilly.com)
    discuss
  25. My Toddler Loves Planes, So I Built Him a Distributed System(jacobstechtavern.com)
    discuss
  26. US DOE Announces $99M, 21 Projects to Advance US Geothermal Energy Development(energy.gov)
    1comments
  27. Informing HTTP Extension Design with Data(mnot.net)
    discuss
  28. Agentic Social Scheduler(github.com/gitroomhq)
    discuss
  29. Signal Will Let You Sign Up Without a Phone Number – For $3(itsfoss.com)
    discuss
  30. We just shipped support for the ugliest part of HTTP: Vary(cloudflare.com)
    discuss

Show HN: DrivingBench – Frontier LLMs Driving a Real Toyota Corolla

3 pointsby 59m agodrivingbench.com
0 comments
Hi all, wanted to share a cool project me and two buddies have been working on. Essentially we took a Toyota Corolla, installed a Comma 4, hooked it up to GPT-6 Astra/Claude Fable/Grok 4.6/Sol, gave them MCP tools to control the Toyota, and then we let them try driving a cone course (in an empty parking lot at low speed). They can observe the cameras/telemetry, request steering/speed, and stop the car. The goal was to see how good frontier LLMs might be just out-of-the-box at a task like driving a real physical car (inspired by recent such demos in robotics).

The best model was GPT-6 Astra which fully finished the course on its second attempt (which was pretty shocking to us). We gave each model up to 3 attempts on the course in the same chat conversation (to incentivize in-context learning). Fable also improved quite a bit from 9% to 45% by its 3rd attempt.

We also have a trace viewer at our website (for example see https://drivingbench.com/trace/gpt-6-astra/2/ for Astra's successful attempt), and everything is open source (traces, harness/code at https://github.com/aditya-ramabadran/drivingbench_harness_v1, we have a report at https://drivingbench.com/report/ on how we did everything, some of the things that went wrong or that we could do better, etc).

Obvious disclaimer: the models clearly aren't good enough to drive on an actual road yet. Also we did this at super low speeds in an open parking lot with a human always ready to brake at any time. Please use the harness/code/etc at your own risk. Would love to answer any questions or take any feedback for v2!

A quiet thread, for now.Start the conversation on HN ↗