JetBrains' Junie Local hits 55 tokens per second on Nvidia RTX Spark
Original title (Chinese)
JetBrains 的本地 AI 编程工具 Junie Local 已 Day 0 适配英伟达 RTX Spark：吞吐最高每秒 55 个词元
AISummary
JetBrains says its local AI coding tool Junie Local is Day 0 compatible with Nvidia's RTX Spark platform, reaching generation speeds of up to 55 tokens per second in preliminary tests.
The Windows version has moved from experimental to stable, and the tool needs no registration, subscription, or token fees.
It also runs on RTX GPUs with at least 24GB VRAM, such as the RTX 3090, 4090, and 5090 with 32GB, and users can choose Qwen3.6-27B or a JetBrains fusion model.
Source: IThome · AI · ithome.comPublished · added here