OreoLook · not peer reviewed
A Three-Layer Caching Architecture for Low-Latency LLM Web Search on Commodity CPU Hardware
OreoLook presents a three-layer cache for session context, semantically equivalent queries, and URL embeddings in an open LLM answer engine operating on commodity CPU infrastructure.
- Research status
- preprint
- Lifecycle
- active
- Relationship
- Connected work
- Registry state
- verified
- Reproducibility
- partial
- Repository license
- CC0-1.0
- Released
- 2026-08-12
Research context
Scope and public claim
OreoLook, formerly lixSearch, is an open answer engine for live web research and cited reports. Its connected public record includes the operated search site, the Elixpo/OreoLook Hugging Face Space, the CC BY 4.0 evaluation dataset, the externally stewarded source repository, and arXiv preprint 2609.05463 on three-layer caching for low-latency LLM web search.
Connected work. Listed because it has a documented technical or publication connection to Elixpo, without being represented as Elixpo-owned.
Reproducibility
partial
The implementation, interactive Space, paper, and evaluation dataset are public; the arXiv record states that code availability is by request, so the complete paper-specific reproduction path is not yet documented.
Review statement
not peer reviewed
Elixpo reports the review state explicitly and does not imply external peer review where none is recorded.
Runnable demonstrations
Inspect the work interactively.
Licensing
evaluation dataset
CC BY 4.0
Repository-level evidence
The externally stewarded source repository currently exposes CC0-1.0. Its maintainers and artifact-specific records remain authoritative.
License checked 2026-10-17 · registry status checked 2026-10-16
Citation
Bhattacharya and Gazi. A Three-Layer Caching Architecture for Low-Latency LLM Web Search on Commodity CPU Hardware. arXiv:2609.05463 (2026).