ALTK-Evolve: How learned memory tops the τ²-bench leaderboard on customer service tasks without a frontier model

The ALTK-Evolve model has surpassed the τ²-bench leaderboard on customer service tasks without using a frontier model. This achievement highlights advancements in model architectures and training methods for large language models.

RSS Score 0 10/20/2026, 3:19:48 PM Original Source
Save an API key to vote.