AFM-4.5B gets Intel CPU and GPU support through OpenVINO 2025.3
- Source
- x.com
- Date
We're happy to announce that AFM-4.5B is fully supported by the latest @intel OpenVINO 2025.3 release (https://t.co/pHxQCCuIKr) You can now easily optimize and run AFM-4.5B on a wide range of Intel CPUs and GPUs, from devices to cloud instances. If you'd like to quickly test, 16-bit, 8-bit, and 4-bit versions for OpenVINO are available on Hugging Face at https://t.co/oRv38ifCw5 ➡️ AFM-4.5 launch blog post: https://t.co/4n6W8NQSdA ➡️ AFM-4.5B (original model): https://t.co/hTs1QUI5MB -- Follow @arcee_ai for the latest news on small language models.
Context
Intel's OpenVINO 2025.3 release, a toolkit for optimizing and running AI models on Intel hardware, added full support for Arcee's AFM-4.5B, following Arcee's earlier work benchmarking the same model across Intel Xeon, AWS Graviton, and Qualcomm CPUs.
Arcee published ready-to-use 16-bit, 8-bit, and 4-bit versions of AFM-4.5B on Hugging Face for OpenVINO, so developers can run or further optimize the model on a range of Intel CPUs and GPUs, from edge devices to cloud instances, without converting the weights themselves first.
- quantization — Shrinking a model by storing its numbers less precisely — like rounding — so it runs faster and fits on smaller hardware, at a small quality cost.
Checking sign-in…
Loading comments…