Will the best LLM in 2027 have <500 billion parameters?

Will the best LLM in 2026 have <500 billion parameters?

Which LLM has more parameters?

More than 80% of all user queries to LLMs will be served by LLMs less than 10 Billion parameters in size by 2050?

When will the first quadrillion parameter LLM be made?

Will the highest-scoring LLM on Dec 31, 2026 show <10% improvement over 2025's best average benchmark performance?

Will an LLM improve its own ability along some important metric well beyond the best trained LLMs before 2026?

Will there by a major breakthrough in LLM continual learning before 2027?

Will any LLM be able to multiply together arbitrary decimal numbers by the end of 2027?

Will there be a state-of-the-art LLM that is NOT based on next raw token prediction before 2029?