JetBrains has announced that its local AI coding tool, Junie Local, now supports the Nvidia RTX Spark platform from day zero. Additionally, the company has transitioned its Windows support from an experimental phase to a stable release.
The tool allows users to read code, modify files, and run tests directly on their local machines. JetBrains emphasizes that the software does not require registration or subscriptions, imposes no token fees, and does not set usage quotas. Furthermore, all code remains on the user's device without cloud uploads.
Performance tests on the Nvidia RTX Spark platform show that Junie Local can achieve generation speeds of up to 55 tokens per second. The installation process automatically verifies hardware and drivers, deploys models, and optimizes context capacity based on available GPU memory.
Hardware requirements include an Nvidia RTX graphics card with at least 24GB of VRAM, based on the Ampere architecture or newer. Supported models include the RTX 3090, 4090, and the 5090 with 32GB of VRAM. Users also need Windows 11, Nvidia driver version 581 or later, and 20GB of free disk space.
The software provides users with a choice of models, including Qwen3.6-27B or a proprietary fused model provided by JetBrains. The installer dynamically adjusts context capacity to ensure the AI retains as much code and conversation history as possible.
