Irvine, California—Razer said on April 30 that it has released an updated version of Razer AIKit, its free, open-source AI development toolkit. The latest release extends the toolkit beyond text-based workflows to support image, video and audio AI models, while adding compatibility with Arm64 architectures.
The company first introduced AIKit at CES 2026. The toolkit is designed to let developers build, run and deploy AI models on local hardware rather than relying entirely on cloud-based services. It handles GPU discovery, configuration and performance tuning, giving developers more control over deployment environments and operating costs.
With the update, AIKit now supports image-generation models including Tongyi-MAI/Z-Image-Turbo and FLUX.2-klein-base-4B. Razer said the same workflow can be used from experimentation through production, including on supported local and edge systems such as recent Razer Blade laptops.
The release also adds support for Arm64-based systems, including platforms using NVIDIA DGX Spark, Grace Hopper and Grace Blackwell Superchips. This could allow teams working across x86 and Arm-based infrastructure to use the same deployment process without reworking applications for each environment.
Razer also said AIKit was used to support AVA Mini, a 3D AI pet companion campaign launched in March 2026 as part of the company’s April Fool’s campaign. Users uploaded photos of real pets and received AI-generated AVA Mini characters.
The campaign ran from March 31 to April 4, 2026. According to Razer, it generated more than 11,000 images, with an average end-to-end turnaround time of 3.24 seconds per image and peak throughput of 30 images per minute. The company said the campaign required no manual intervention during the five-day deployment.
To run the campaign, Razer partnered with Akash Network, a decentralized GPU marketplace. AIKit was deployed through AkashML, Akash’s managed inference service, across a distributed pool of RTX 4090 and RTX 5090 GPUs. Razer said this lowered inference costs to about US$0.01 per image, compared with typical cloud API pricing of US$0.03 to US$0.15 per image.
The companies have also published a joint white paper outlining the deployment architecture, cost structure and engineering decisions behind the campaign.
Razer AIKit is now available through Akash Console, Akash Network’s browser-based interface for deploying and managing applications. This gives developers and GPU operators a way to deploy AIKit workloads without using command-line tools.
Razer said future releases will add voice and video capabilities, extending the toolkit’s use cases to voice-driven interaction and video-generation workflows. AIKit is currently available on GitHub.
