dom96

Joined 16 years ago

Karma

7,397

Submissions

2,225

Member since

Jan 2

Currently working as a Software Engineer at Cloudflare on Workers (and particularly on Python Workers). Previously at Meta.

Since 2010 I have been involved with building the Nim programming language, as well as many tools and libraries for it. I have also published a book about Nim called Nim in Action.

These days I'm working on various personal projects- some big, some small. Most recently https://onlyhumanhub.com, https://listifications.app and https://mousetrack.co.uk

Github: http://github.com/dom96

Bluesky: https://bsky.app/profile/dom96.picheta.me

Twitter: http://twitter.com/d0m96

Mastodon: https://mastodon.social/@dom96

Website: http://picheta.me, my contact info is here, always happy to receive emails.

View on Hacker News

Recent submissions

Comment · 3 days ago

Will this make it to OpenRouter? Running your own benchmarks against these models is surprisingly frustrating when you don’t want to create hundreds of accounts for each AI lab’s API.

Comment · 7 days ago

not yet, but it will be, won't it?

Comment · 8 days ago

2 runs for each task, for each task it's either a pass/fail (the small green/red dots show each task pass/fail)

Comment · 8 days ago

I'd say it's both. Models should be intelligent enough to be able to tell prompt injections apart. But the language itself is also novel and not in the model training data set, so that should be a good signal for intelligence in itself.

Comment · 8 days ago

Glad you like it. Ideas/contributions welcome on GitHub btw https://github.com/dom96/killswitch

Comment · 8 days ago

yeah, it performed best for whatever reason. More specifically, it managed to deduce the meaning of the hack statement in the language better than Opus 5.5 did.

Comment · 8 days ago

A lot of these models are likely benchmaxxing the existing benchmarks that are out there. Some perhaps even by accident.

I think looking at some more niche benchmarks helps, I created one and it does show Kimi K3.1 Pro as the most intelligent of all open-weight models, though it is still significantly below Opus 5/5.5[1]

1 - https://bench.killswitch-lang.org/

Comment · 8 days ago

> beating both GPT-6 Astra and Fable 5.1. It's also faster than both.

Very similar to my adversarial programming language (KillSwitch) benchmark[1], at least on pass rate, it is much slower though.

1 - https://bench.killswitch-lang.org

Comment · 8 days ago

Ran it on my adversarial programming language (KillSwitch) benchmark[1]. Very significant improvement over K3. On par with Gemini 4 Argon and Fable 5.1. Much slower than both though.

But this is now the highest scoring open-weight model. Amazing stuff.

1 - https://bench.killswitch-lang.org

Comment · 8 days ago

So I just created a brand new Kimi API account and it seems I have no access to this new model. Is this expected?

Comment · 10 days ago

Has anything changed? I already ran it on my adversarial programming language benchmark[1] a few weeks ago and it performed pretty poorly (on par with Kimi K3).

1 - https://bench.killswitch-lang.org/

Comment · 12 days ago

For what it's worth, it does perform on par with DeepSeek V4 Pro on my adversarial programming language benchmark[1]. I wouldn't think they would benchmax on that.

1 - https://bench.killswitch-lang.org/

Comment · 12 days ago

Yes, humans can learn which parts of the spec are true and which aren’t. LLMs can too, but only through their context window (or long term through retraining)

Comment · 12 days ago

They can use tools. The benchmarks run in a VM and run a coding agent which has permissions to do whatever it wants (except access the internet).

Comment · 12 days ago

There is a reference implementation: https://github.com/dom96/KillSwitch.

In the benchmarks the models/coding agents get a binary of the interpreter and indeed some of them do reverse engineer it. Over time this will get into their training data. But we can evolve the language and evade their training as we go.