undisclosed

taxonomy axis: openness

Decision rule: Report is silent on release. Absence is itself the classification.

Anchor: First qualifying report sets the anchor

Models in this category (6)

ModelOrgDateBlockAttentionContextTotalActive
GPT-2OpenAInot discloseddense1,0241.542B1.542B
Phi-3Microsoft2024-04-22denseMHA4,0963.8B3.8B
Phi-4Microsoft Research2024-12-12denseMHA4,09614B14B
Qwen3.5Qwen Team2026-04-21hybridhybrid262,144not disclosednot disclosed
ZAYA1Zyphra2026-05-06sparse-MoEGQA131,0728.4B760M
VibeThinkerWeibo2026-06-15densenot disclosed65,5363B3B

What varies within this category

attention variant varies — GQA, MHA, hybrid, not disclosed. block type varies — dense, hybrid, sparse-MoE. trained context varies — 1024, 131072, 262144, 4096, 65536. scale class varies — frontier, medium.