Run Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s
Strata claims to run 125B Qwen 3.8 Flash Next on an RTX 4090 at about 100 tokens per second.
A Hacker News post links to the GitHub project Strata, which claims to run the 125-billion-parameter Qwen 3.8 Flash Next model on a consumer NVIDIA RTX 4090 at about 100 tokens per second. The thread has 48 points and 13 comments. No independent benchmark, model weights release, or security issue is described in the listing.
46