AI News

Edge0 open-sources 145B on-device model Edge0-145B and 120B version

Hacker News TrendingUpdated Nov 13 09:57

AI summary

Edge0 has open-sourced two on-device models, Edge0-145B and Edge0-120B, and simultaneously released a unified inference framework and deployment tools. The company says the 145B version achieves 29GB peak memory, 80+ tokens/s decoding speed, and 1100+ tokens/s prefill on compatible laptops, supports 128K–256K context, and scores 64.6 on DeepSWE 1.1, 86.52 on Terminal-Bench 2.1, and 90.4 on GPQA-Diamond. The model weights have been uploaded to Hugging Face, the inference framework is open-sourced on GitHub, and an offline-usable Edge0 App is also provided.

Why read this

SelectedAI score 85

The official results provide measured data for the 145B model on a laptop—29GB peak memory and 80+ tokens/s—along with comparisons across six benchmarks, which can be used to assess the practical limits of running large models on-device.

Topics

Tags

Body

Only a summary is authorized here. Continue at the validated source link.

Source

Hacker News Trending