Third Conference on Language Modeling

The Third Conference on Language Modeling (COLM 2026) is a dedicated venue for research on language models—their architecture, training, evaluation, capabilities, limitations, and societal implications. Emerging from the recognition that language modeling has grown into a discipline of its own, COLM provides a focused forum where the full depth and breadth of language model research can be examined without being subsumed under a larger AI conference program.

Researchers in deep learning, NLP, AI safety, computational linguistics, and AI ethics attend to present work on scaling, alignment, interpretability, emergent capabilities, multilingual modeling, and benchmark design. COLM 2026 is equally interested in critical and skeptical perspectives on language model claims as in positive results, making it a valuable corrective to hype-driven discourse. It is essential for anyone whose research centers on understanding what language models actually do and why.