Limited-time GPU firepower Dirt-cheap LLM Inference: Llama 4, DeepSeek 0528

Comments (1)

dikobraz · 8h ago

We’ve got a temporarily underutilized 64 x AMD MI300X cluster, so instead of letting it sit idle, we’re opening it up for LLM inference.

Running: LLaMA 4 Maverick, DeepSeek V3, R1, and R1-0528. Want another open model? Let us know. We are happy to deploy it.

Prices are around 50% lower than the cheapest OpenRouter endpoints, and they will stay that way through June (maybe July).

The server handles up to 10,000 requests/sec, and we allocate GPUs per model based on demand. So feel free to load-test it, hammer it, or run production traffic. We're collecting no data whatsoever.

cloudrift.ai/inference

Full disclosure: I am the founder. We're trying to make good use of this capacity. Let us know if you have any ideas on how to utilize the cluster meaningfully. We're happy to hear the feedback.

Self-hosted x86 back end is now default in debug mode (ziglang.org)

Why Android can't use CDC Ethernet (2023) (jordemort.dev)

Tracking Copilot vs. Codex vs. Cursor vs. Devin PR Performance (aavetis.github.io)

Riding high in Germany on the world's oldest suspended railway (theguardian.com)

FSE meets the FBI (blog.freespeechextremist.com)

Omnimax (computer.rip)

Administering immunotherapy in the morning seems to matter. Why? (owlposting.com)

The wire that transforms much of Manhattan into one big, symbolic home (2017) (atlasobscura.com)

Panjandrum: The 'giant firework' built to break Hitler's Atlantic Wall (bbc.com)

Endangered classic Mac plastic color returns as 3D-printer filament (arstechnica.com)

Kagi Reaches 50k Users (kagi.com)

Forests offset warming more than thought: study (news.ucr.edu)

Analyzing IPv4 Trades with Gnuplot (ipv4a-5539ad.gitlab.io)

My first attempt at iOS app development (mgx.me)

Gaussian integration is cool (rohangautam.github.io)

Web designs are getting too complicated (websmith.studio)

Show HN: Let’s Bend – Open-Source Harmonica Bending Trainer (letsbend.de)

Software Is About Promises (bramadams.dev)

Joining Apple Computer (2018) (folklore.org)

Enterprises are getting stuck in AI pilot hell, say Chatterbox Labs execs (theregister.com)

Generating Pixels One by One (tunahansalih.github.io)

Self-Host and Tech Independence: The Joy of Building Your Own (ssp.sh)

The last six months in LLMs, illustrated by pelicans on bicycles (simonwillison.net)

Building supercomputers for autocrats probably isn't good for democracy (helentoner.substack.com)

How Compiler Explorer Works in 2025 (xania.org)

How to get started with writing tech video essays

FAA to eliminate floppy disks used in air traffic control systems (tomshardware.com)

<Blink> and <Marquee> (2020) (danq.me)

Building an AI server on a budget (informationga.in)

Binfmtc – binfmt_misc C scripting interface (netfort.gr.jp)

Poison everywhere: No output from your MCP server is safe (cyberark.com)

Launching the BeOS on Hitachi Flora Prius Systems (1999) (testou.free.fr)

Cheap yet ultrapure titanium might enable widespread use in industry (2024) (phys.org)

Coventry Very Light Rail (coventry.gov.uk)

Field Notes from Shipping Real Code with Claude (diwank.space)

Efficient mRNA delivery to resting T cells to reverse HIV latency (nature.com)

OpenBSD IO Benchmarking: How Many Jobs Are Worth It? (rsadowski.de)

Focus and Context and LLMs (taras.glek.net)

There's not much point in buying Commodore (oldvcr.blogspot.com)

tcpulse: A TCP/UDP load generator that provides fine-grained, flow-level control (github.com)

Rkyv (archive) is a zero-copy deserialization framework for Rust (rkyv.org)

Ask HN: How to learn CUDA to professional level

Low-Level Optimization with Zig (alloc.dev)

Why We're Moving on from Nix (blog.railway.com)

The time bomb in the tax code that's fueling mass tech layoffs (qz.com)

What was Radiant AI, anyway? (blog.paavo.me)

I used AI-powered calorie counting apps, and they were even worse than expected (lifehacker.com)

A tool for burning visible pictures on a compact disc surface (2022) (github.com)

Startup Equity 101 (quarter--mile.com)

A look at Cloudflare's AI-coded OAuth library (neilmadden.blog)

Limited-time GPU firepower Dirt-cheap LLM Inference: Llama 4, DeepSeek 0528

Comments (1)