Source · Tier A · Peer-reviewedSource · zkGPT: An Efficient Non-interactive Zero-knowledge Proof Framework for LLM Inference
zkGPT: An Efficient Non-interactive Zero-knowledge Proof Framework for LLM Inference
W. Qu, Y. Sun, X. Liu, T. Lu, Y. Guo, K. Chen, J. Zhang. 2025. 34th USENIX Security Symposium (USENIX Security 25), pp. 2045–2063.
| Original | https://www.usenix.org/conference/usenixsecurity25/presentation/qu-zkgpt |
|---|---|
| Version | Proceedings paper read from the USENIX PDF (usenixsecurity25-qu-zkgpt.pdf). Code archived on Zenodo, DOI 10.5281/zenodo.14727819 (v1, 23 January 2025, CC BY 4.0). |
| Accessed | 2026-09-25 |
| Note | National University of Singapore and Hong Kong University of Science and Technology. Proves GPT-2 inference with GKR, Lasso lookups and Hyrax commitments, made non-interactive with Fiat–Shamir; Table 3 reports 21.8 s proving (32 threads), 0.35 s verification and a 101 KB proof on a 16-core Xeon server with 200 GB of memory. |