crewAI 1.15.27: New XPU Support, Deepinfra Provider, and Improved Evaluation Tracking
crewAI version 1.15.27 introduces XPU support for the OpenCLIP provider, adds Deepinfra as an OpenAI-compatible provider, and includes enhancements around evaluation tracking and cost reporting.
What changed?
crewAI version 1.15.27 adds several notable features: XPU has been added to the device options for the OpenCLIP provider, and Deepinfra is now available as an OpenAI-compatible provider. The update also improves evaluation and deployment tracking by recording why evaluations stop, what a deployment attempt met, and adding per-run cost and time tracking for model evaluations. Additionally, bug fixes address stop sequences sent to some OpenAI models and improve logging.
Why does it matter to an everyday developer?
The new XPU support expands hardware options for running OpenCLIP, potentially improving performance or compatibility for users with relevant hardware. Deepinfra's inclusion as an OpenAI-compatible provider enables switching or expanding to a new cloud AI provider with minimal changes, increasing deployment flexibility. Enhanced evaluation logging and per-run cost/time reporting improve transparency and make it easier to analyze and optimize resource usage during model evaluation, which is valuable for both development and operational monitoring.
What can the developer do now?
Developers using crewAI can update to version 1.15.27 to benefit from new provider options and more informative evaluation tracking. Users interested in running OpenCLIP with XPU hardware or evaluating Deepinfra as a provider can now do so directly. Developers should also consider leveraging the enhanced cost and evaluation tracking for better monitoring and optimization of their AI workflows.
