A new large language model, Ling-3.1-flash, has been released by Chinese labs. This model boasts approximately 560 billion total parameters with 25 billion active parameters per token and supports a context window of up to 1 million tokens. It has demonstrated strong performance across various domains, achieving an Elo score of 1,673 on GDPVal-AA v2.1, 75.16 on FrontierSWE, and 65.35 on HealthBench Professional. Similar to previous releases, the model is available for free use for two weeks before becoming open source. AI
IMPACT Sets a new benchmark for context window size and performance in open-source LLMs.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →