frontier LLMs

and resistance to prompt injection attacks. Improving instruction hierarchy in frontier LLMs

social interaction topics, validated by native speakers. Our evaluation of five frontier LLMs reveals substantial gaps in cultural competence, with accuracy rates 40-48% below native speakers when taarof is culturally appropriate. Performan

struct, and (ii) agentic approaches including zero- and few-shot prompting with frontier LLMs (LLaMA-3.3-70B-Instruct and GPT-4o) and a Mixture-of-Agents (MoA) framework that leverages a diverse set of LLMs by combining outputs from multi-l

. We explore two forms of pseudo feedback based on test cases: one generated by frontier LLMs and the other by extending self-consistency to multi-test-case. We conduct experiments on both mathematical reasoning and coding tasks using pseud

s Gemini as a family of highly capable multimodal models. Gemini belongs to the frontier LLMs timeline after December 2023.

c introduced Claude on March 14, 2023 as an AI assistant. Claude belongs to the frontier LLMs timeline after March 2023.

vember 30, 2022 as a conversational AI research release. ChatGPT belongs to the frontier LLMs timeline after November 2022.