frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Show HN: Self-host open-source LLMs on AWS with scale-to-zero

https://github.com/paguasmar/veloxml-deploy
3•paguasmar•3h ago

Comments

paguasmar•3h ago
Basically the CLI

1. compiles your model code and dependencies without requiring Dockerfiles or K8s

2. provisions spot or on-demand instances directly in your own AWS/GCP account (via SkyPilot)

3. it then spins up the runtime and gives you a production-ready URL endpoint

I wrote this originally to replace Modal and Baseten, which I wasn't too pleased with. I needed to deploy open-source models inside my own VPC without vendor lock-in or proprietary Python decorators.

Thought the community here might like it. AWS is an amazing cloud - with one-command deployment and scale-to-zero I was able to consolidate all my LLM serving into my own infrastructure for a fraction of the cost.

Would love feedback. There are a few (many?) bugs and a lot of things to iron out, but I'm working with some close friends to make it awesome.

dsemakin•3h ago
You got me interested, a rough table of model size to instance type to spot $/hr would help a lot. The 0.5B example is CPU only, so it doesn't say much about what a 7B or 70B actually costs.
paguasmar•2h ago
You're right. A CPU only model is definitely is not a real prod workload where you'd need GPU machines.

Here is a rough table of model size to instance type to spot $/hr:

For up to 8B, you can use a c6i.xlearge which costs as spot about $0.4/hr.

For up to 70B, you can use a g5.12xlearge that costs you about $2/hr.

Besides the GPU machine be aware that you need a controller instance to route the requests and scale. That has a fixed cost of up to $0.08/hr. When not in use at all, just type veloxml down --all and it tears down the controller too for true $0/hr.

We're currently running larger LLM models benchmarks to add to the README this week.

Show HN: Copperhead – Cursor for circuit boards

https://copperhead.sh/
226•animeshchouhan•18h ago•90 comments

Show HN: LLM Attention Visualization

https://ishamf.dev/p/llm-attention-visualizer/
155•ifz•14h ago•24 comments

Show HN: ZWPlayer 3.3.2, a free HTML5 video player with ads and music mode

https://www.zwplayer.com/
2•logicplayer•1h ago•0 comments

Show HN: AI-first open-source alternative to Microsoft and Google Office

https://github.com/injectinglabs/injoffice
3•nickpismenkov•2h ago•3 comments

Show HN: Estimate your AI CO2 footprint

https://llmfootprint.fyi/
4•wyre•2h ago•0 comments

Show HN: VolAnti – Open-source acoustic detector for fibre-optic FPV drones

https://github.com/agamrossen/VolAnti
8•agamrossen•7h ago•0 comments

Show HN: Self-host open-source LLMs on AWS with scale-to-zero

https://github.com/paguasmar/veloxml-deploy
3•paguasmar•3h ago•3 comments

Show HN: Raven – Encrypted messages hidden inside ordinary text

https://sendraven.ink
5•willhieb•4h ago•0 comments

Show HN: Jigsaw Haiku

https://jigsawhaiku.com/
110•windowshopping•4d ago•33 comments

Show HN: Interactive Tree of Life

https://ptree.org/
107•Lucent•5d ago•25 comments

Show HN: Full feature ECS game engine for TypeScript

https://github.com/oasys-works/oecs
2•2khan•6h ago•0 comments

Show HN: SimpleW – Lightweight C# Web Server with optional addons

https://simplew.net/
4•prodbro•10h ago•0 comments

Show HN: Browser-based Nginx config generator with live preview

https://tools.techtransit.org/nginx-generator
4•techgpt•7h ago•0 comments

Show HN: Stuxnet – A reconstructed source code of the infamous cyber-weapon

https://github.com/Sadpainy/Stuxnet
169•CMDDestory•1d ago•50 comments

Show HN: DriveSync – fast Git-styled Google Drive sync CLI

https://github.com/scaleninja/drivesync
18•rohityadavcloud•18h ago•12 comments

Show HN: Making a GBA game with GPT-6 Astra

https://www.spritefusion.com/blog/making-a-game-boy-advance-game-with-gpt-6-astra
8•HugoDz•14h ago•1 comments

Show HN: Pipnote – Share a reminder they can't peek at until it arrives

https://pipnote.app/
4•radomird•11h ago•3 comments

Show HN: Auto-Rig Web – Zero-click 3D humanoid rigging in a Web Worker

https://github.com/nff747/auto-rig-web
2•nff747•8h ago•0 comments

Show HN: GET Together – A social network where you don't need POST to Post

https://gettogether.dev
106•nchudleigh•2d ago•59 comments

Show HN: Genotype: language that compiles to idiomatic TS, Rust and Python

https://genotype-lang.org/
7•kossnocorp•9h ago•0 comments

Show HN: InverSQL: Build SQL interactively in inverse

https://github.com/rentruewang/inversql
4•renchuw•14h ago•0 comments

Show HN: Wg-admin – web UI for an existing WireGuard host

https://github.com/logimaxx/wg-admin
32•vsergione•1d ago•11 comments

Show HN: Browser viz of OpenAI's "spaghetti" Navier–Stokes vortex (illustration)

https://3d-retro.com/experiments/vortex
3•SouthWestAtlas•11h ago•0 comments

Show HN: Mador – Make any DOM reactive with a tiny 80-line Proxy state tuple

https://github.com/marsbos/mador
100•bosmarcel•2d ago•34 comments

Show HN: Rokz: a browser port of a 1990 ASCII dungeon crawler

https://taler.pl/rokz/
2•htaler•11h ago•1 comments

Show HN: A browser IDE for m68000, Z80 assembly (and more), with graphics

https://asm-editor.specy.app/
6•specy•20h ago•1 comments

Show HN: Engrim – A universal, local-first SQLite memory engine for AI CLIs

https://github.com/timgordontg/engrim
90•timgordontg•2d ago•53 comments

Show HN: Linubot – like Grok bot and Hermes had a kid

https://github.com/agent-sh/linubot
3•anotherCodder•12h ago•0 comments

Show HN: Pen – A writing tool where conversation is the editor

https://pen.xyz
3•harleyverse•13h ago•0 comments

Show HN: HomeCat – Design your backyard office

https://myhomecat.com
39•dearilos•1d ago•24 comments