HyperAccel

About HyperAccel

HyperAccel is a Seoul semiconductor company building LPU (Language Processing Unit) hardware and software for LLM inference. Products include Bertha accelerator cards, a Forte accelerator and an HX server.

Common use cases

• LLM inference serving
• Edge LLM deployment
• Inference cost reduction

Products and services

Key features

• LLM inference silicon
• vLLM and Hugging Face support
• PyTorch support
• Server platform
• IP licensing

Integrations

• vLLM
• Hugging Face
• PyTorch

HyperAccel alternatives

• GroqLPU inference chips and the GroqCloud inference service
• TenstorrentRISC-V based AI processors, accelerator cards and IP licensing
• RebellionsSouth Korean AI inference chip and accelerator maker
• FUFuriosaAI

FAQs