Hacker News: comandillos

New comment by comandillos in "GLM-5.3 is now open-weight"

comandillos — Sat, 29 Aug 2026 09:36:04 +0000

I mean, I think it depends. At home 3 of us we use AI for multiple reasons, from coding apps to asking general questions, and if we would have to pay equivalent subscriptions that would be ~1k a year on AI + submitting all your data to external services. I payed around ~8k on 2 DGX Sparks that, at the moment, serves perfectly fine as a ChatGPT/Claude replacement at home (DS4 Flash peaking at ~170 tokens per sec with 6 concurrent sequences), and even once the technology is obsolete for inference in a few years, I will still have 2 pretty powerful machines for whatever I need + some pretty fast NVME Storage. I don't think its a terribly bad idea.

New comment by comandillos in "RAG Is Simpler Than You Think"

comandillos — Sat, 29 Aug 2026 09:09:22 +0000

Not really, I just made a really simple Python library the agent can use and modify as he wishes within a sandboxed environment to explore the SQLite database. When the user prompts the agent, the prompt goes into my coding agent, omp, and starts running Python scripts and throwing SQL sentences until coming up with the answer. It works extremely well in our experience.

New comment by comandillos in "RAG Is Simpler Than You Think"

comandillos — Thu, 27 Aug 2026 16:47:33 +0000

Its a read-only SQLite file. And I mean the people using the chatbot knows it uses AI so just like Google they shouldnt pick the first result, but forcing the LLM to mention the sources and not assume acronyms works amazingly well

New comment by comandillos in "RAG Is Simpler Than You Think"

comandillos — Thu, 27 Aug 2026 06:30:48 +0000

I have a web user interface connected to a coding agent (OMP) running inside a container, so if any of the tool calls fail or something happens, usually my model recovers autonomously from these situations. The capabilities of models like DS4 Flash are those of frontier models from months ago, so its recovery and autonomous capabilities are quite impressive.

New comment by comandillos in "RAG Is Simpler Than You Think"

comandillos — Wed, 26 Aug 2026 20:27:42 +0000

I indexed thousands of documents into a SQLite database with an FTS5 index, plugged it into DeepSeek v4 Flash and got better much better results than any other commercial solutions my company has tried in the past.

The trick was just to let the LLM come up with its own SQL queries for searching... and the results are impressive.

New comment by comandillos in "GLM-5.3-Flash"

comandillos — Wed, 26 Aug 2026 18:11:27 +0000

https://github.com/bethington/ghidra-mcp . Works flawlessly.

New comment by comandillos in "GLM-5.3-Flash"

comandillos — Wed, 26 Aug 2026 18:07:52 +0000

I've been reverse engineering LEON3-FT SPARC v8 BE code, so I wouldn't say it's common :D. When attached to Ghidra through a MCP the things you can do with this are simply crazy.

New comment by comandillos in "GLM-5.3-Flash"

comandillos — Wed, 26 Aug 2026 17:55:46 +0000

I am surprised. I've been using DS4 Flash (0731) for weeks now and it works perfectly fine as a replacement for Claude in a large variety of cases. It requires a few more iterations, sure, but it's useful enough to not need a Claude subscription anymore. Among the things I do I've been reverse engineering, writing complex C++ code...

New comment by comandillos in "The AI Situation in Software Development"

comandillos — Sat, 15 Aug 2026 16:19:56 +0000

Sure if you use remote AI services, but any companies working on niche markets where they want to protect their IP, or they simply work with sensitive stuff, will rely on local AI instead.

New comment by comandillos in "AI At Home Part 1: A Box Of Scraps"

comandillos — Thu, 13 Aug 2026 19:11:12 +0000

Absolutely agreed, I don't see this as an investment leading to cost-savings in the future, at all.

I see this rather as an investment in improving my capabilities and knowledge of this technology, letting me play with a 'GPU cluster', vLLM and other technologies that otherwise would require me to rent GPUs on the cloud.

New comment by comandillos in "AI At Home Part 1: A Box Of Scraps"

comandillos — Thu, 13 Aug 2026 18:39:48 +0000

I ended up buying 2 DGX Sparks interconnected over QSFP with the intention of getting rid (as much as I could) of any cloud-based AI provider. I'm running DS4 Flash 0731 on it and some OCR models, using Oh My Pi and OpenWebUI as my main ways to interface with the agent... and from someone that has been using Claude for a long time, I can definitely say I don't need it anymore. Not for the stuff I'm doing.

New comment by comandillos in "DeepSeek V4 Pro 0813"

comandillos — Wed, 12 Aug 2026 19:22:42 +0000

At least you can run it for relatively cheap hardware. I guess OpenAI doesn't let you do that.

New comment by comandillos in "I'm switching my phone from Android to Linux"

comandillos — Sat, 08 Aug 2026 10:32:02 +0000

I'm pretty hyped about OpenHarmony.

I've been using HarmonyOS for almost a year now (not the AOSP fork, but the actual Hongmeng microkernel flavour)... and this idea of having a iOS-like system with an Android VM running at near native performance works pretty well. Unfortunately is really focused on the chinese market but it feels really polished and promising.

New comment by comandillos in "DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis"

comandillos — Fri, 31 Jul 2026 18:28:33 +0000

Here's you pelican just in case you were guessing

https://imgur.com/a/j4s2dt3

New comment by comandillos in "Kuna: Decompiler Development in the Age of Coding Agents"

comandillos — Thu, 30 Jul 2026 13:07:05 +0000

Haven't used IDA much lately, but after looking at the screenshot with that IDA PRO decompiled code in their website I feel like Ghidra is already ahead of them in this area :D

New comment by comandillos in "Moonshot AI suspends new subscriptions due to Kimi K3 demand"

comandillos — Sun, 19 Jul 2026 18:34:25 +0000

Oh, so they’ve decided to suspend subscriptions because they know they might not be able to guarantee a minimum quality of service to their customers?

New comment by comandillos in "Qwen 3.8"

comandillos — Sun, 19 Jul 2026 09:10:04 +0000

Just imagine Anthropic making Opus open-weights now for the sake of trolling everyone. Wouldn't surprise me at this point xD

New comment by comandillos in "Bonsai 27B: A 27B-Class model that runs on a phone"

comandillos — Tue, 14 Jul 2026 20:33:54 +0000

Quite weird that heavy quantization method on a dense model gives better results than slightly quantized MoE models like 35B-A3B from Google.

At this point all the different quantization and 'compression' (look at MPO applied to LLMs...) techniques start feeling a bit like snake oil. It's just gut feeling - or scores on benchmarks models are optimized for - what ends up deciding whether a technique is good enough or not.

New comment by comandillos in "Physical disc production ending in Jan 2028 for new games on PlayStation"

comandillos — Wed, 01 Jul 2026 16:52:50 +0000

It backfired with the PSP Go. It will backfire again. No-go I would buy a console without disks. Sorry. No.

New comment by comandillos in "Nvidia is proposing a beast of a CPU system for Windows PCs"

comandillos — Sat, 06 Jun 2026 22:56:49 +0000

I dont really get the hype with all the N1X thing when in reality this is the same almost 1 yr old GB10 that was released with the DGX Spark and proved to be quite a disappointment