Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Have You Seen Shannon(news.ycombinator.com ↗)
    discuss
  2. AI Lab building tycoon game(bedroomlabs.net ↗)
    discuss
  3. An Undercover Google Analyst Infiltrated a Notorious Supply-Chain Hacking Gang(wired.com ↗)
    discuss
  4. Before you build it, share it(coidea.co ↗)
    discuss
  5. I vibecoded a tool to shrink Golang codebase automatically(github.com/dhilst ↗)
    1comments
  6. SSD-Llama: SSD-Native Inference for Trillion-Parameter Moe on a Consumer PC(arxiv.org ↗)
    discuss
  7. ChatGPT's Built-in Browser now supports Chrome extensions(twitter.com/jameszmsun ↗)
    discuss
  8. Keeping vLLM's Prefix Cache Warm Between Agent Turns(doug.sh ↗)
    discuss
  9. Near-real-time vision using DiffusionGemma in Jev mode(twitter.com/mmastrac ↗)
    discuss
  10. Anthropic finally adds AGENTS.md support to Claude Code(twitter.com/trq212 ↗)
    1comments
  11. The Highway Tutorial(fastcode.substack.com ↗)
    1comments
  12. Show HN: Iceoryx2 0.10: flatbuffer integration, zerocopy IPC with unbounded data(ekxide.io ↗)
    discuss
  13. Show HN: Open-source revenue recognition and analytics for Stripe and Metronome
    discuss
  14. I built a project database for data science agents(yizhao95.github.io ↗)
    discuss
  15. Open Challenge List(openchallengelist.com ↗)
    1comments
  16. Ask HN: So is AI taking our jobs?
    1comments
  17. Flawed benchmarks (epoch.ai benchmark registry filtered by flawed)(epoch.ai ↗)
    discuss
  18. Anthropic builds wet lab for AI-driven drug discovery(qz.com ↗)
    1comments
  19. Show HN: I compiled FreeRDP into a Flutter app so my phone can do RDP and SSH(xconnectapp.com ↗)
    discuss
  20. WeirdML v3(htihle.github.io ↗)
    discuss
  21. Show HN: SIM Screener adds device frames to screenshots and recordings(simscreener.com ↗)
    discuss
  22. Show HN: Semantic-skill-kit:agent skills adapted from Google Modern Web Guidance(github.com/ulrischa ↗)
    discuss
  23. Border agents can search cellphones without a warrant or reasonable suspicion(lawandcrime.com ↗)
    3comments
  24. Europe's AI firms, playing catch-up, challenge US calls for slowdown(reuters.com ↗)
    discuss
  25. Lmjtfy – Ask Jev a yes or no question(lmjtfy.dev ↗)
    1comments
  26. China's CAS institute charts DUV route to 3nm GAA without EUV(digitimes.com ↗)
    discuss
  27. Stop waiting to live (before AGI decides for you)(youtube.com ↗)
    discuss
  28. 2026 Global Cities Index(oxfordeconomics.com ↗)
    discuss
  29. TypeSafe / Jev latency-focused demos built by Devin(github.com/dabit3 ↗)
    discuss
  30. Show HN: Turnpanel is a local first AI workspace(turnpanel.com ↗)
    discuss

We made Playwright 2x faster and 80% more token efficient

14 pointsby 2h agogithub.com
14 comments
2h agoHN ↗

We built Stagehand 2 years ago (24k stars and 4M monthly npm downloads) and recently fixed its biggest flaw: round-trip latency.

Every action performed requires a round trip between your script and the browser (short when running locally but increased when running in the cloud). We also saw multiple posts complaining about the eager token appetite of Playwright MCP.

For this reason, we rebuilt Stagehand from the ground up and shipped v4, where Stagehand controls the browser from an extension automatically loaded upon your browser startup.

Stagehand v4 comes with batch command support, dedicated token-efficient methods `act()` and `extract()`, and a brand new architecture making it 2x faster than Playwright and 80% more token efficient.

You can see for yourself by looking at our benchmarks, comparing its performance across a dozen models (frontier and open weights) and tools (Codex, Claude Code, and more): https://www.stagehand.dev/evals

Ask me anything!

1h agoHN ↗

what did you lose compared to CDP (e.g. cross-origin iframes, downloads running in envs where you can't load extensions)?

1h agoHN ↗

It still uses CDP but communicates from an extension within the browser instead of a script running in a separate runtime or, worse, in a separate region.

1h agoHN ↗

I tried this myself - my conclusion was the overhead of parsing a screenshot, generating an action, and being limited to headful mode is much less efficient than reading accessibility trees & generating CDP commands.

basically, browser automation is a closer-to-the-metal abstraction than computer use, allows more flexibility, and ends up being much cheaper at scale!

1h agoHN ↗

Astra with computer use seems to burn a ton of tokens though. Stagehand seems more token efficient.

1h agoHN ↗

How much does the harness really matter for evals?

1h agoHN ↗

A lot, especially for performance. That's why we built our own benchmarks that account for both the model and the harness. For example, with Claude Opus 5, the accuracy gap can be up to 3% and performance up to 200ms, depending on whether you're using Deep Agents, Eve, or Fx.

More details here: https://www.stagehand.dev/evals

1h agoHN ↗

So for an enterprise with a 1,000 test Playwright suite, does this basically mean ~2x faster CI times? That would be huge.

1h agoHN ↗

Exactly, as Stagehand now runs inside the browser, you'll save on the round trip. Also, enabling batch actions will further accelerate your test suite.

1h agoHN ↗

Do I need to pay for Browserbase to use this?

1h agoHN ↗

Nope, Stagehand is open-source and works with local browsers by default.

38m agoHN ↗

seems really cool! although, one question i have is why not keep agent() alongside the new primatives? it seems v4 removed agent() entirely rather than offering it with all of act(), observe(), extract().

32m agoHN ↗

We removed agent because so many great harnesses are available in the ecosystem. Instead of keeping it, we decided to make Stagehand v4 better integrated with popular harnesses, both at the Coding Agent level (Codex, Claude Code) and frameworks level (Eve, Deep Agents, Mastra, etc)