FastFlowLM Developers Join AMD To Help Push Open-Source NPU Software
([AMD] 5 Hours Ago
FastFlowLM + AMD)
- Reference: 0001646976
- News link: https://www.phoronix.com/news/FastFlowLM-Joins-AMD
- Source link:
On top of releasing [1]ROCm 7.14 as the new production release of ROCm now built offTheRock, rolling out the [2]Lemonade 11.0 local AI server, and [3]GAIA 0.22 , AMD has some more open-source news in the lead up to next week's AMD Advancing AI event.
AMD announced today that the developers behind the FastFlowLM open-source developers have joined the company. In an [4]AMD blog post today, they confirmed FastFlowLM developers have joined the company:
"Recently, the FastFlowLM team joined AMD, marking another key step in our strategy to advance AI performance and efficiency across the stack. FastFlowLM developed a lightweight, highly optimized inference software flow that delivers fast, efficient large language and multimodal model performance directly on AMD technology-powered AI PCs and workstations."
FastFlowLM has long been focused on the AMD Ryzen AI NPU for acceleration in an ollama-style developer experience that runs on Windows. It's been one of the premier ways to actually utilize the AMD Ryzen AI NPU. Until today's announcement, I didn't realize that FastFlowLM developers weren't employed by AMD... I had assumed so given their long focus on Ryzen AI and close relation, but now they are formally part of AMD. The FastFlowLM company was started by academy researchers for advancing AI/LLMs on parallel NPU architectures.
"We remain committed to investing in this open ecosystem, and we’re thrilled to build the future of on-device AI together."
FastFlowLM is already open-source and so jives nicely with the rest of the open-source AMD AI efforts from ROCm to Lemonade, GAIA, and more. The developers are joining the AMD AI Group to focus on client and workstation AI software and day zero enablement of the latest models.
The FastFlowLM code is available under an MIT license on [5]GitHub .
[1] https://www.phoronix.com/news/AMD-ROCm-7.14
[2] https://www.phoronix.com/news/AMD-Lemonade-11.0
[3] https://www.phoronix.com/news/AMD-GAIA-0.22
[4] https://www.amd.com/en/blogs/2026/fastflowlm-joins-amd-to-advance-ai-inference.html
[5] https://github.com/FastFlowLM/FastFlowLM
AMD announced today that the developers behind the FastFlowLM open-source developers have joined the company. In an [4]AMD blog post today, they confirmed FastFlowLM developers have joined the company:
"Recently, the FastFlowLM team joined AMD, marking another key step in our strategy to advance AI performance and efficiency across the stack. FastFlowLM developed a lightweight, highly optimized inference software flow that delivers fast, efficient large language and multimodal model performance directly on AMD technology-powered AI PCs and workstations."
FastFlowLM has long been focused on the AMD Ryzen AI NPU for acceleration in an ollama-style developer experience that runs on Windows. It's been one of the premier ways to actually utilize the AMD Ryzen AI NPU. Until today's announcement, I didn't realize that FastFlowLM developers weren't employed by AMD... I had assumed so given their long focus on Ryzen AI and close relation, but now they are formally part of AMD. The FastFlowLM company was started by academy researchers for advancing AI/LLMs on parallel NPU architectures.
"We remain committed to investing in this open ecosystem, and we’re thrilled to build the future of on-device AI together."
FastFlowLM is already open-source and so jives nicely with the rest of the open-source AMD AI efforts from ROCm to Lemonade, GAIA, and more. The developers are joining the AMD AI Group to focus on client and workstation AI software and day zero enablement of the latest models.
The FastFlowLM code is available under an MIT license on [5]GitHub .
[1] https://www.phoronix.com/news/AMD-ROCm-7.14
[2] https://www.phoronix.com/news/AMD-Lemonade-11.0
[3] https://www.phoronix.com/news/AMD-GAIA-0.22
[4] https://www.amd.com/en/blogs/2026/fastflowlm-joins-amd-to-advance-ai-inference.html
[5] https://github.com/FastFlowLM/FastFlowLM