Localmaxxing says it now tracks 3,500+ local AI runs across 444 model setups
LocalMaxxing · reddit · 2026-07-22
A builder says they created localmaxxing, a site for collecting and sharing local AI inference benchmarks.
- The platform now claims roughly 1,860 users, 3,500+ runs, 444 model/quants, and 199 hardware entries.
- It is adding an eval system so users can upload custom evaluations, with traces and stats stored for later analysis.
- It also includes a marketplace for used hardware and a rental system for sharing endpoints.
- The project is built with agents in mind, with API docs and a CLI meant to make integration easy.
- The author pitches it as a centralized place for local inference benchmarks and evals.
More from coding & agent
- Tldraw schedules its first hack night at a new office near Old Street — max__drake · 2026-07-22
- Open-source multAIplayer Alpha makes Codex a shared team coding workspace — jxnlco · 2026-07-22
- Full course on graph engineering shows how to build safer, faster AI agent workflows — EXM7777 · 2026-07-22
- Training an Agent Class 2 adds distillation resources, slides, and recording — SergioPaniego · 2026-07-22
- July 28 training class will cover GRPO and TRL for agent training — SergioPaniego · 2026-07-22
- Karpathy's Tip: Rambling to LLMs via Voice is a Great Prompting Strategy — nptacek · 2026-07-22