Jan 0.8.6 hotfix addresses Linux AppImage and NVIDIA GPU issues

The Jan open source desktop AI assistant has released version 0.8.6 as a hotfix for version 0.8.5. This update resolves model loading crashes in the Linux AppImage by using the system's Vulkan loader instead of a bundled one. It also fixes compatibility issues for NVIDIA Turing GPUs (GTX 16xx, RTX 20xx) on older drivers and improves memory management for Qwen3.8 with speculative decoding.

Technical improvements in Jan 0.8.6

• Linux AppImage: Switched from a bundled Vulkan loader to the system's loader to prevent crashes on various distributions.
• NVIDIA Turing Support: Ships native Turing code to allow CUDA engine functionality on drivers older than R595.
• Memory Management: Fixed a performance regression in Qwen3.8 with MTP; when Parallel Sequences is set to auto, the engine now uses one slot instead of four, preventing models from being pushed into shared system memory on 12 GB cards.
• MCP Tools: Images returned by MCP tools are now passed directly as images rather than base64 text, improving compatibility with vision-capable models.

Why it matters

This update is critical for Linux users utilizing AppImages and owners of older NVIDIA hardware who require stable local AI execution.