Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU
Llama-modchip aims to run a 284-billion-parameter DeepSeek model using 64GB of RAM without a GPU. The provided information does not explain its technical approach. (github.com)
Show HN: Logo Deal – Logos anyone can edit with text
logo.deal9 points3 comments
logo.deal is a self-service AI logo generator built around editable vector output rather than finished image files. Users describe a brand, its purpose, preferred mood and colours, then choose a style such as minimal, playful, luxury or tech. The tool produces logo concepts as SVGs that can be opened in design software, with PNG exports also available. Users can change colours, text and layout themselves at no extra cost, and iterate with AI edits where their plan’s credits allow. The service does not include a human designer. Visitors can try one free generation without signing up; previews are watermarked, and a browser-based template demo offers a separate way to test the workflow without AI.
Paid plans provide clean downloads and commercial-use rights. Starter costs $15 monthly or $144 annually and includes 160 monthly credits; Pro costs $39 monthly or $374 annually with 420 credits; Agency costs $99 monthly or $950 annually with 1,050 credits and brand-kit exports. Annual plans include a seven-day trial, but require a card; trial downloads are watermarked, and billing starts after seven days unless canceled. Monthly plans have no trial. A four-concept quick generation costs five credits, while an AI edit costs ten; manual edits are free. Credits generally reset each billing period, with limited rollover for some extras. Commercial use applies to paid-plan downloads, but users should check for trademark conflicts before registering a design.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
More in AI
Incredible take from one of the GNU maintainers
A GNU maintainer’s controversial take prompted debate about AI-generated code in open-source projects. Commenters discussed project leaders who use AI tools while distancing themselves from code ownership. (social.treehouse.systems)
ThinkStream – AI generates the work. We still need to understand it
ThinkStream says AI can generate work, but people still need to understand it. The provided page content contains only an anti-bot prompt. (thinkstream.app)
I Use Development-Driven Specs, Not Spec-Driven Development
Ensemble is a book and open-source project for building an AI coding agent that helps create the next edition. It advocates development-driven specifications, where specs evolve alongside implementation instead of dictating development upfront. (github.com)
Show HN: Maple – Sub-2-BPW Moe Pruning to Run Gemma 4 26B at 45 TPS on a 6GB GPU
Maple uses manifold-activated pruning to compress mixture-of-experts language models below 2 bits per weight. Its developers say it runs Gemma 4 26B at 45 tokens per second on a 6GB GPU. (github.com)
Goodbye Claude, Hello Pi
A longtime Claude Code user is switching to Pi after growing frustrated with incompatible instruction-file formats, limited VSCode session support, and bugs in the GUI extension. They say Claude Code has prioritized adding features over fixing reliability and usability problems. (caleb.software)