DESILO's THOR selected as reference implementation for encrypted AI in global FHE benchmarking suite, setting a standard for private LLM inference
Executive summary: DESILO Inc. announced that its THOR framework has been selected as the reference implementation for encrypted AI in a newly launched global FHE benchmarking suite. It establishes a standardized way to evaluate and compare the performance of encrypted large language model inference, helping enterprises assess private AI solutions.
Who is involved: DESILO Inc., the THOR framework developers, and the consortium behind the global FHE benchmarking suite.
Likely next: The benchmark is expected to be adopted by AI hardware vendors and cloud providers to validate FHE-enabled LLM deployments, with further updates likely as the suite evolves.
DESILO announced that its THOR framework has been chosen as the reference implementation for encrypted AI within a new global FHE benchmarking suite. The move aims to create a universal benchmark for measuring the performance and security of private large language model inference. By providing a standardized testbed, the initiative could accelerate enterprise adoption of encrypted AI technologies and guide investment in FHE acceleration.
Timeline
- — DESILO's THOR Selected as Reference Implementation for Encrypted AI in Global FHE Benchmarking Suite (PR Newswire)
Analysis — what this means
Sectors affected
- AI infrastructure
- encrypted computing
- FHE hardware