Yhn HealthFitness first seen 18 h ago, last 40 min ago, peak #9

Qwen3.8 hits 255 tokens per second on one RTX 5090

Original: Running Qwen3.8 at 255 tok/s on a single 5090

Engineers are discussing running the Qwen3.8 model at 255 tokens per second on a single Nvidia RTX 5090 GPU. The result, shared by meshapplied.com, shows how far consumer-grade hardware has come for local large language model inference, with commenters weighing in on the optimization techniques and what it means for running capable models without data-center equipment.

Why now: It demonstrates that current consumer GPUs can serve modern LLMs at near-instant speeds, a notable milestone for local inference.

Qwen3.8Nvidia RTX 5090MeshApplied

Open on hn →

Rank over time, top of the chart is #1. 14 snapshots from 18 h ago to 40 min ago.

Evidence

• Running Qwen3.8 at 255 tok/s on a single 5090 · Ir0n · 14

API: https://socialmediatrends-api.osmike.com/v1/trends/2164021