Zyphra Hierarchical Attention (ZHA) is a tree-indexed sparse attention method that retrofits frozen LLMs with 4B tokens of distillation, matching dense attention on reasoning and 256K retrieval benchmarks. The post Zyph…
New models reset the capability and price-performance frontier. Teams re-evaluate what to build on whenever a launch shifts what's possible per dollar.
Summaries are aggregated for information only — follow the source link for the full story. Demo entries are illustrative.