AMD's GAIA Defaults To Better Model, Continued Improvements For Local AI
AMD has released version 0.17.5 of its open-source GAIA platform, which simplifies running local AI agents on AMD hardware across Windows and Linux. The update replaces the previous default Qwen 3.5 35B model with Google's Gemma 4 E4B, offering improved multimodal capabilities and performance. While GPU utilization was confirmed, the Ryzen AI NPU was not engaged during testing, indicating room for further optimization.
- ▪AMD GAIA 0.17.5 now uses Gemma 4 E4B as the default model for both language and vision tasks.
- ▪Gemma 4 E4B is a natively multimodal model with ~4.5B effective parameters and a 128K context window.
- ▪The update adds OpenAI tool_calls support, a lightweight chat agent, and semantic code search via CodeAgent.
- ▪The agent UI is now bundled in the PyPi wheel, simplifying installation via 'pip install amd-gaia[ui]'.
- ▪During testing, the Ryzen AI NPU showed 0% utilization despite compatible software and hardware.
Opening excerpt (first ~120 words) tap to expand
AMD's GAIA Defaults To Better Model, Continued Improvements For Local AI Written by Michael Larabel in AMD on 2 May 2026 at 04:19 PM EDT. 5 Comments AMD software engineers on Friday released a new version of GAIA "Generative AI Is Awesome" as their open-source software for Windows and Linux leveraging the Lemonade SDK and aiming to make it easy to build AI agents on your PC with all local AI processing across AMD's CPUs, GPUs, and NPUs. AMD GAIA continues steadily improving and making it easier to get local AI agents running on your AMD hardware whether it's on Windows or Linux. With GAIA 0.17.5, they have replaced Qwen 3.5 35B with Gemma 4 E4B as the default model.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Phoronix.