A publication dated October 15, 2025 describes Recursive Language Models, which interact recursively with very long prompts through a REPL, and reports results on the OOLONG and BrowseComp-Plus benchmarks.
The publication, dated October 15, 2025, presents Recursive Language Models (RLMs) as an inference strategy for decomposing and recursively interacting with very long prompts through a REPL. It reports results on the OOLONG and BrowseComp-Plus benchmarks, but the available summary provides no specific metrics or comparisons.
The approach is presented as a possible alternative to explicit retrieval for tasks involving long contexts. To assess the proposal, consult the original publication and check its test setup, full results, and stated limitations; do not infer performance beyond what the published evidence supports.