Hacker News: bhu8

New comment by bhu8 in "OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?"

bhu8 — Sun, 25 Oct 2026 22:12:33 +0000

It’s probably more efficient caching methods and smaller models (6 sol is actually 6 terra and opus 5.5 is smaller than 5.1 etc). Don’t overthink this please

New comment by bhu8 in "Show HN: agent-usage-guard – a Claude Code mod to protect your usage window"

bhu8 — Thu, 22 Oct 2026 18:22:48 +0000

You could use a dumber model (luna or haiku) for compact to get around this

New comment by bhu8 in "A Statistical Analysis of LLM Benchmaxxing"

bhu8 — Fri, 16 Oct 2026 17:35:25 +0000

tl;dr: Google stands out with statistical significance; Alibaba (Qwen) is probably in second place, but data are not conclusive.

A Statistical Analysis of LLM Benchmaxxing

bhu8 — Fri, 16 Oct 2026 17:35:25 +0000

Article URL: https://llemre.com/benchmaxxing/

Comments URL: https://news.ycombinator.com/item?id=50121608

Points: 1

# Comments: 1

New comment by bhu8 in "Sharing AI progress in mathematics"

bhu8 — Wed, 07 Oct 2026 04:38:20 +0000

The applied mathematician’s joke is that log n is bounded above by 45 or so.

New comment by bhu8 in "ChatGPT Pro 500"

bhu8 — Tue, 29 Sep 2026 18:28:54 +0000

> I can't imagine OpenAI not backtracking this unless Anthropic does the price hike too.

Same here, but I’m worried it’s the latter. We’ll probably see an Anthropic price hike in a couple of weeks.

New comment by bhu8 in "Opus 5.5 is good at explainer videos"

bhu8 — Fri, 25 Sep 2026 18:21:17 +0000

These are garbage but Opus 5.5 can actually do much better: https://x.com/emollick/status/2103272686570918334

New comment by bhu8 in "MCP was always a bad idea?"

bhu8 — Mon, 21 Sep 2026 17:39:25 +0000

For anything with CLI access and permissions CLI >>> MCP, not even close

Also this how most ADEs enable agent messaging and the like

New comment by bhu8 in "Google is making private AI practical with homomorphic encryption"

bhu8 — Fri, 14 Aug 2026 20:30:45 +0000

I wouldn't be very bullish.

Homomorphic encryption got significantly efficient with the first few iterations, but I don't really see the necessary orders of magnitude savings coming soon. You could reduce this by some partial encryption schemes (e.g., for LLMs you need a handful of basic operations) but a better alternative already exists: multi-party computation.

Source: I did research in this area in the past.

New comment by bhu8 in "GPT-5.6, Grok 4.5, Claude, and Muse Spark build the same 4 apps"

bhu8 — Fri, 10 Jul 2026 21:58:59 +0000

Absolutely yes, but that's how you become twitter/X famous

New comment by bhu8 in "Herdr: Agent multiplexer that lives in your terminal"

bhu8 — Mon, 29 Jun 2026 17:58:48 +0000

May I plugin my own stuff? https://getviberia.com

New comment by bhu8 in "Ask HN: What are you working on? (June 2026)"

bhu8 — Sun, 14 Jun 2026 21:28:27 +0000

Factorio/SimCity like interface for managing multiple agents: https://getviberia.com

It's like the love child of Polytopia and Conductor. As many other agent management platforms/harnesses, Viberia has been building itself, and honestly this has been too much fun to stop.

New comment by bhu8 in "Anthropic's model naming, extrapolated"

bhu8 — Wed, 10 Jun 2026 19:58:41 +0000

I like how the original triplet's initials represent their behavior well:

- Opus is OP, like OverPowered

- Sonnet is SO, like your significant other (this was more meaningful in Sonnet 3.7 days)

- Haiku is HA, like the reaction to a bad joke

The latest model, naturally, needed three letters: FAB.

I'm now looking forward to ABS and LO

New comment by bhu8 in "Open source Kanban desktop app that runs parallel agents on every card"

bhu8 — Sat, 23 May 2026 05:02:08 +0000

Just use direnv? You’ll probably need to adjust the port you are hosting the local page on, but that’s just N=mod(hash based on the worktree name) and then port=default_port+N.

Tell your claude to set this up. Should do it in a single prompt

Show HN: Viberia – Civ/Polytopia-like command center for AI agents (BYOK/BYOS)

bhu8 — Tue, 19 May 2026 08:02:28 +0000

Hey HN,

This is my take on the agent harness. Everything on an isometric map. Agents are grouped into "buildings" that run in a sequence or a loop; e.g., the CodeForge has an agent that writes a PRD, another one that implements, and a third that reviews. Everything is customizable, you build your own buildings/teams however you want.

It's a Tauri app, really light (about 8x less energy than the closest competitor I benchmarked, so it actually runs from a coffee shop on battery). It's macOS only for now, but ping me if you are willing to test the Windows or Linux version.

I've been dogfooding this for months and would love to get some feedback, feature requests, and bug reports so I know what to focus on next.

Comments URL: https://news.ycombinator.com/item?id=48190531

Points: 1

# Comments: 0

New comment by bhu8 in "New Claude Code programmatic usage restrictions"

bhu8 — Wed, 13 May 2026 21:58:26 +0000

It would unfortunately impact it. ACP uses Claude SDK and is developed by a third-party.

New comment by bhu8 in "GPT-5.5"

bhu8 — Thu, 23 Apr 2026 20:53:50 +0000

Gpt-5.3-codex is miles better than 5.4 in that regard. It’s better at orchestration, and does the things that it said it did. Haven’t tested 5.5 yet but using 5.4 for exploration + brainstorming and handing over the findings to 5.3-codex works pretty well

New comment by bhu8 in "Elevated errors on Claude.ai, API, Claude Code"

bhu8 — Wed, 15 Apr 2026 16:42:39 +0000

Feels like an issue in their caching. First non-cached turns are sent properly but everything that is second+ turn fails.

New comment by bhu8 in "Ask HN: What Are You Working On? (April 2026)"

bhu8 — Mon, 13 Apr 2026 16:50:39 +0000

I am working on a (yet another) local app for managing multiple claude/codex/gemini sessions in a game like environment: https://getviberia.com/

New comment by bhu8 in "Show HN: The Lottery of Life"

bhu8 — Wed, 18 Mar 2026 07:19:06 +0000

I think something is broken though. I got 20 nematodes in a row. It's around 1% prob.