CrowdGPT: a crowd-trained, datacenterless 1B open-source LLM, 6.9B tokens done
Vxtzq1 · reddit · 2026-10-25
A developer launched CrowdGPT, an effort to build the first datacenterless 1B-parameter LLM in a fully open-source, collaborative way — anyone can contribute compute, inspect the code, and steer the project. The claim: you don't need massive GPU clusters or millions of dollars to train a commercial-grade model. The model has already been trained on 6.9B tokens using the framework itself, and the author is soliciting testers, code reviewers, and critics (crowdgpt.net, GitHub).
More from Infra
- Epoch AI: Memory Chips Shipped Through 2027 Could Run 33–171 Million Frontier AI Agents — PeterDiamandis · 2026-10-25
- AI Studio's 1M-token counter is an illusion: how cold cache and rereads really burn your quota — Former-Aerie6530 · 2026-10-25
- Is a €250 modded Dell T5810 + RTX 3090 a smart local-LLM host in 2026? — SlechteConcentratie · 2026-10-25
- Apples-to-apples rerun: DeepInfer tops DeepSeek's published throughput by ~9% on 4x B300 — charles_irl · 2026-10-25
- Apollo's viral "circular AI spending" charts debunked: Ramp card data skews to seed startups — 2C_ornot2C · 2026-10-25
- Open source and local AI as a counterbalance is dawning on people — mjdramstead · 2026-10-25