Understand this story
On October 22, 2026, MemTensor open-sourced its native memory architecture model MemT-1.0, offering two size tiers, 36B Nano and 288B Flash, and released a 66-page technical report. According to representative reports, the model places a Native Memory Block in parallel alongside the standard Transformer Block, enabling token-level dynamic state updates through erase, write, and forget gates, and expands memory capacity via a Mixture-of-Storage mechanism. Data cited in the reports states that with 256K of history, VRAM drops from 5.06 GiB to 40 MiB, and first-token latency remains at 104–112 ms between 512 and 256K. There are currently no subsequent developments.
AI-generated from public reports
Latest development · Report
MemTensor Open-Sources Native Memory Model MemT-1.0, Available in 36B and 288B TiersDevelopments
1 confirmed developments
MemTensor open-sources native memory architecture model MemT-1.0
Source:Ayi AI Notes
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 22
AYi AI Notes
MemTensor Open-Sources Native Memory Model MemT-1.0, Available in 36B and 288B Tiers
The MemTensor team has open-sourced MemT-1.0, a model built on a native memory architecture, offering two size tiers—36B Nano and 288B Flash—and has released a 66-page technical report. The model connects a Native Memory Block in parallel alongside the standard Transformer Block, uses an erase gate, write gate, and forget gate to achieve token-level dynamic state updates, and extends memory capacity through a Mixture-of-Storage mechanism. According to the figures in the article, under a 256K history, VRAM drops from 5.06 GiB to 40 MiB, and first-token latency remains at 104–112 ms between 512 and 256K.