AuroraIMG2-50M-Preview: a 50M-denoiser text-to-image model that runs on a 2016 laptop CPU
Tall_Abrocoma_3533 · reddit · 2026-10-28
AuroraAI-Research released AuroraIMG2-50M-Preview, a tiny text-to-image model with a 50M-parameter denoiser and 90M end-to-end inference parameters including a custom text encoder and pretrained Flux VAE. It generates images in 27s at 50 steps (11s at 20 steps) on a 2016 Intel i5-7200U laptop CPU. Licensed under Apache 2.0.
Related event: AuroraIMG2-50M: Tiny Text-to-Image Model Runs on Old CPUs(2 posts)→
More from Multimodal
- Mundo turns photos of real places into explorable, editable interactive worlds — Scobleizer · 2026-10-28
- Google's new AI world model reportedly simulates entire worlds in real time — heykahn · 2026-10-28
- NVIDIA open-sources Nemotron 3.5 Transcribe 30B ASR model; fine-tuning cuts word error rate 45.8% — NVIDIAAI · 2026-10-28
- NVIDIA open-sources Nemotron 3.5 Transcribe: 30B MoE ASR model with ~4B active params — NVIDIAAI · 2026-10-28
- Agentic video editing is 'solved': hundreds of videos/day, timeline editing obsolete — ZeroStateReflex · 2026-10-28
- Seedance 2.0 Mini demo: bubble-trick penguin AI wildlife clip — Burak_Antakya · 2026-10-28