Once the USA wakes up, this will be insane news

What's special about Claude Code? Isn't Opus the real magic?

Surely there's nothing here of value compared to the weights except for UX and orchestration?

Couldn't this have just been decompiled anyhow?

I think pi has stolen the top honors, but people consider the Claude code harness very good (at least, better than Cursor)

Pi is the best choice for experts and power users, which is not most people.

Claude Code is still the dominant (I didn't say best) agentic harness by a wide margin I think.

Pi really is amazing. It's as much or as little as you need it to be.

Not having to deal with Boris Cherny's UX choices for CC is the cherry on top.

Already made the switch back to Codex :-)

Help us out with Terminal Bench 3.0!

https://docs.google.com/document/d/1pe_gEbhVDgORtYsQv4Dyml8u...

It uses Claude Code natively. If you're still neurotic about it, you can also use opencode with gastown.

Are there credible sources for these claims?

These type of stories are typically attributed to anonymous sources. So it’s impossible for a reader to conclude anything about the sources’ own credibility. So they usually rely on the credibility of the reporter or media org who’s published about it.

The Information is also reporting on this, but paywalled. IMO The Information does solid reporting.

https://www.theinformation.com/newsletters/applied-ai/openai...

Editing to add some of the Information’s reporting: > Speaking at a panel at Davos moderated by The Information CEO Jessica Lessin, Friar suggested that in the field of drug discovery, her company could, for instance, take a “license to the drug that is discovered” using OpenAI’s technology. In other words, OpenAI would take a profit-sharing stake in the financial upside its AI creates for customers.

> Friar is no doubt familiar with older AI drug discovery firms such as Recursion that struck deals with pharmaceutical firms to give them big bounties for successful drugs identified by their tech. There aren’t many, if any, examples of such successes yet, though.

> OpenAI isn’t the only firm eyeing this opportunity. Its rivals Anthropic, Google DeepMind and Isomorphic Labs, an Alphabet subsidiary focusing on using AI for drug discovery, have also held discussions with early-stage biotechnology startups about data licensing or partnerships.

We have the same insight into the sources as we do OpenAI’s finances.

A rigorous study of the "unreasonable effectiveness" method.

Why Johnny can't be unreasonably effective?

What we talk about when we talk about the unreasonable effectiveness

Reasonable effectiveness: an alternative to unreasonable effectiveness

Does anyone understand what the end objective of V is? I've known about this language for I think 4-5 years and for some reason it's still v0.5...?

> Does anyone understand what the end objective of V is?

The Vlang site makes it clear what their objectives are[1]. "Simple, fast, safe, compiled. For developing maintainable software." They also have a roadmap[2], showing what their end objective version (or production release) will have.

> I've known about this language for I think 4-5 years and for some reason it's still v0.5...

They just came out with version 0.5 (2 weeks ago from this post), so obviously that statement is mistaken.

[1] vlang.io

[2] https://github.com/vlang/v/blob/master/ROADMAP.md (vlang's roadmap)

No it has not been 0.5 for years. Check their homepage. You can see the timeline for the releases there. 0.5 was released a few days ago and is a major step for the language.

There is a noprocrast feature in your settings to specify how long you can stay on for a single session and the frequency at which you can view HN. Super helpful!

There is a noprocrast feature in your settings to specify how long you can stay on for a single session and the frequency at which you can view HN. Super helpful!

Never forget Suchir Balaji.