Hacker News: Bolwin

New comment by Bolwin in "Sam Altman may control our future – can he be trusted?"

Bolwin — Tue, 07 Apr 2026 04:33:28 +0000

I find it bizarre even the public image of Anthropic is seen as ethical after the Department of War debacle, in which they themselves admitted they had basically no qualms with their tech being used for war and slaughter at all except two very very thin lines, namely mass surveillance of American citizens and fully automated weaponry with their current models.

It only showed they were marginally more ethical than OpenAI and XAI which isn't saying much.

New comment by Bolwin in "Cohere Transcribe: Speech Recognition"

Bolwin — Tue, 31 Mar 2026 21:25:17 +0000

Try two newlines between each one

New comment by Bolwin in "Cohere Transcribe: Speech Recognition"

Bolwin — Tue, 31 Mar 2026 21:24:09 +0000

Many ASR models already support prompts/adding your own terminology. This one doesn't, but full LLMs especially such expensive ones aren't needed for that.

New comment by Bolwin in "Openrouter Going Rogue?"

Bolwin — Sun, 29 Mar 2026 17:29:27 +0000

From their discord:

-----

For those experiencing 403 errors when accessing certain models:

We're required to comply with the terms of service of our upstream model providers. Our enforcement mechanisms and regional access rules are updated continuously. We understand that enforcement changes are disruptive for people who didn't realize their usage was in violation. You can find the relevant policy in the Prohibited Content section of our [Terms of Service](<https://openrouter.ai/terms#_6_-prohibited-conduct_>).

This is not a ban on using OpenRouter. We will keep working hard to add all the best models so everyone has great options to choose from, no matter where you are.

If you are seeing a 403 author banned error and believe you are not in violation of the Terms of Service of the AI Model or Provider you are using, please fill out the [appeal form](<https://forms.gle/yc2vyJiALz8Uhbmh7>) with detailed information. We will review these and correct any mistaken restrictions.

New comment by Bolwin in "Chroma Context-1: Training a Self-Editing Search Agent"

Bolwin — Sat, 28 Mar 2026 20:31:08 +0000

What does kimi do? Can you point me to a link?

New comment by Bolwin in "Claude loses its >99% uptime in Q1 2026"

Bolwin — Sat, 28 Mar 2026 20:13:26 +0000

I'd say claude is whole foods

New comment by Bolwin in "Why Sora Failed: $15M/day inference cost vs. $2.1M lifetime revenue"

Bolwin — Thu, 26 Mar 2026 10:02:38 +0000

This page reloads infinitely for me, can't see it

New comment by Bolwin in "Thoughts on slowing the fuck down"

Bolwin — Wed, 25 Mar 2026 21:51:16 +0000

It's interesting how everything you list is created problems in the tools themselves.

If you step back and just look at "can this do what I wanted" without worrying about what shit storm of software makes it work.

Mind you perfectionists will always have work. That doesn't mean anything.

New comment by Bolwin in "TurboQuant: Redefining AI efficiency with extreme compression"

Bolwin — Wed, 25 Mar 2026 16:10:31 +0000

Do you know if this also applies to the muon optimizer? It seems to be replacing adamw

New comment by Bolwin in "Verilator open-source SystemVerilog simulator"

Bolwin — Sat, 21 Mar 2026 23:38:52 +0000

I don't know a ton about this field (though I'm curious, if anyone had recommended material) but I used Verilator for a risc-v project in university and it was excellent, substantially faster than Yosys

New comment by Bolwin in "Grace Hopper's Revenge"

Bolwin — Tue, 17 Mar 2026 10:29:30 +0000

That benchmark is useless for comparing languages because the tasks are not the same across languages

New comment by Bolwin in "1M context is now generally available for Opus 4.6 and Sonnet 4.6"

Bolwin — Sat, 14 Mar 2026 06:12:12 +0000

How many big refactorings are you doing? And why?

New comment by Bolwin in "Chrome extension adjusts video speed based on how fast the speaker is talking"

Bolwin — Fri, 13 Mar 2026 12:56:27 +0000

Very interesting idea, but doesn't seem to work very well. I tried on three videos. On two it miscounted to an absurdly low rate of ~1 syllable/s and played it at the max speed which was illegible and in one it failed to detect any speech at all.

New comment by Bolwin in "Are LLM merge rates not getting better?"

Bolwin — Thu, 12 Mar 2026 22:59:51 +0000

I've found Haiku to be truly mediocre for working with. If you want a cheap models, the open source ones are much better

New comment by Bolwin in "No, it doesn't cost Anthropic $5k per Claude Code user"

Bolwin — Tue, 10 Mar 2026 05:57:29 +0000

Yeah that's a massive assumption they're making. I remember musk revealed Grok was multiple trillion parameters. I find it likely Opus is larger.

I'm sure Anthropic is making money off the API but I highly doubt it's 90% profit margins.

New comment by Bolwin in "Just-bash: Bash for Agents"

Bolwin — Thu, 26 Feb 2026 17:29:07 +0000

I've had LLMs write some pretty complex powershell on the fly. Still a shell language but a lot nicer.

Ideally something like nushell but they don't know that well

New comment by Bolwin in "Code Mode: give agents an API in 1k tokens"

Bolwin — Tue, 24 Feb 2026 06:08:41 +0000

I don't know, for practical use, providing a search tool seems more practical than a whole js sandbox only for them to just run text search 99% of the time anyway

New comment by Bolwin in "Anthropic officially bans using subscription auth for third party use"

Bolwin — Mon, 23 Feb 2026 18:45:31 +0000

They still have that. If that's all you wanted then what are you complaining about

New comment by Bolwin in "Keybee: A Keyboard Designed for Smartphones"

Bolwin — Mon, 23 Feb 2026 14:08:36 +0000

Looks like a copy of https://play.google.com/store/apps/details?id=ch.icoaching.t.... I don't know which one can't first.

I tried typewise. Decided it wasn't really worth it.

New comment by Bolwin in "The path to ubiquitous AI (17k tokens/sec)"

Bolwin — Sat, 21 Feb 2026 16:29:38 +0000

Nanbeige. Yeah this seems ideal for models that scale test time compute