All topics

LLM stories

Large language models have moved from research curiosity to production infrastructure. This topic focuses on LLM APIs, fine-tuning, RAG pipelines, agent frameworks, and the economics of running models at scale.

HN discussions here tend to be unusually practical — benchmark comparisons, cost breakdowns, and real deployment war stories from teams shipping LLM features.

898 stories archived · Page 24 of 30

RSS feed for LLM

Arbitrary file read and remote code execution in Active Storage

discuss.rubyonrails.org · 5 points · 0 comments

Launch HN: Tokenless (YC S26) – Automatic model switching to save money

usetokenless.com · 71 points · 63 comments

The Scientific Literature Is Poisonous to LLMs

reinvent.science · 26 points · 11 comments

Extra Headroom – cut Claude Code and Codex token costs by ~50%

extraheadroom.com · 2 points · 0 comments

Escaping the LLM Coding Rat Race

aaronstannard.com · 5 points · 1 comments

NightRun, bare metal LLM inference, no OS, boots from USB

github.com · 6 points · 0 comments

P2Present – slides and talk video in sync, preserved on any storage

p2present.com · 3 points · 2 comments

An Android and iOS app from 7 Claude Code commands, every prompt and timing

proandroiddev.com · 5 points · 0 comments

RC Setlist – Setlist manager and lyric prompter for Ableton Live 12

github.com · 2 points · 0 comments

Multi-agent LLM editor with local inference via WebSockets

x-agent.sascha10k.biz · 2 points · 0 comments

Agent and RAG for Obsidian, Need Feedback

2 points · 1 comments

TokenTown: A visual way to understand how LLMs work

laurentiugabriel.github.io · 74 points · 21 comments

Burnless – 1,590 tokens of context for a 1.44M-token workday

github.com · 2 points · 1 comments

JungOcean – Free personality test with no email gate and local storage

jungocean.com · 4 points · 0 comments

CodeCrucible: A blueprint for LLM-driven SAST

engineering.block.xyz · 4 points · 0 comments

Vyne – Zapier for DeFi users, run from any LLM or the workflow builder

app.vyne.finance · 2 points · 0 comments

I sent Claude Opus 5 '–-' and it wrote me 5k tokens about a cartographer

austinsnerdythings.com · 16 points · 6 comments

OpenReviewer: A Specialized LLM for Generating Critical Scientific Paper Reviews

aclanthology.org · 12 points · 0 comments

What's Wrong with American Studies?

radicallypragmatic.org · 9 points · 5 comments

What if useful AI is a fantasy?

lzon.ca · 28 points · 61 comments

Palette Manipulator

bztsrc.gitlab.io · 3 points · 8 comments

Writekin – fine-tune a local LLM on your own writing, on your Mac

github.com · 5 points · 3 comments

"Uncensored" open LLMs are measurably more optimistic than their base models

arxiv.org · 43 points · 22 comments

SOTA on the hardest AI memory benchmark (BEAM, 10M tokens), with a smaller model

2 points · 3 comments

Now is the time to give LLMs access to the ACM digital library

cacm.acm.org · 190 points · 178 comments

Ctrlb-decompose: Strip the noise from logs before sending to LLMs

github.com · 48 points · 6 comments

Claude Code system prompt says not to ask for permission, assumes user is absent

elliotmilco.substack.com · 4 points · 0 comments

Mondragon Corporation – a federation of co-operatives

en.wikipedia.org · 174 points · 40 comments

DMARC has been public since 2012 but most company domains still don't enforce it

ciphercue.com · 200 points · 168 comments

Does Speaking to Agents Like Cavemen Save 65% of Tokens? We Test

blog.jetbrains.com · 43 points · 41 comments