hotai.news
← Back to AI Newsflash
AI Briefings / Content Guide
AIHOT · Published on 2026-10-04 22:15

Strata open-sourced: running Qwen3.8-Flash-Next (125B) at about 100 token/s on consumer GPUs like RTX 4090

AI Brief

Strata is free and open-source (MIT license), allowing Qwen3.8-Flash-Next (125B) to run on ordinary gaming PCs with over 12 GB VRAM, reaching about 100 token/s generation speed on RTX 4090, with the model about 70 GB download, requiring over 32 GB RAM and about 80 GB disk.

Return to original text

A short introduction is retained here; the full text from the source website is not reposted. For specific wording and subsequent updates, please refer to the original article.

Open AIHOT original