@AdinaYakup: Metis Memory Foundation Model released by Memtensor Research Group Probably the first LLM that internalizes memory into…
Summary
Memtensor Research Group released Metis, a family of LLMs (4B/9B/27B) that internalize memory into the backbone, eliminating external RAG. The model performs memory read/write in a single forward pass and deploys with frozen weights like a standard LLM.
View Cached Full Text
Cached at: 07/31/26, 06:53 PM
Metis Memory Foundation Model released by Memtensor Research Group
Probably the first LLM that internalizes memory into its backbone
- 4B/9B/27B weights + code + paper
- No external RAG: memory is inside the model
- One forward pass = full read/write
- Weights frozen at inference: deploys like a standard LLM
Similar Articles
Metis: Memory Foundation Model
The paper introduces memory foundation models, with Metis as the first prototype, which equips foundation models with native memory capabilities through a new architecture and training data.
@dair_ai: // Memory as a Model // The paper augments any LLM with a separate trained memory model that stores, retrieves, and int…
MeMo introduces a modular memory model that augments any LLM to store, retrieve, and integrate new knowledge without retraining or catastrophic forgetting. It outperforms RAG-based methods on benchmarks like BrowseComp-Plus, NarrativeQA, and MuSiQue.
SimpleMem: Efficient Lifelong Memory for LLM Agents
Introduces SimpleMem, an efficient memory framework for LLM agents that uses semantic lossless compression to improve accuracy and reduce token consumption, achieving 26.4% F1 improvement and up to 30x reduction in inference-time token usage.
Understanding Is Done Early: A Depth Division of Labor in Large Language Models and Its Use for Unbounded-Context Memory
This paper introduces CoMem, a method that exploits the depth-wise division of labor in LLMs to cache intermediate residual tensors and recompute only upper layers for retrieval, enabling bounded read compute and memory independent of stored-context length. Evaluated on Qwen3-8B, CoMem achieves strong long-context performance with significant memory savings and prefill speedups.
MemDefrag: Latent Memory Defragmentation for Large Language Models
Proposes MemDefrag, a training-free framework that uses a middle-layer tracing signal to defragment latent memory in LLMs, achieving significantly better knowledge retention than existing methods like MemoryLLM and M+.