AI/ ai-research · model-architecture · llm-efficiency · arxiv

Researchers Split AI Memory From Reasoning to Speed Inference

A new architecture called Mobius separates knowledge storage from reasoning, letting a 35B model run inference nearly 4x faster at similar accuracy.

A new model architecture separates what an AI knows from how it reasons, and the split makes inference dramatically cheaper.

Researchers describe Mobius-v0, an architecture built around a single shared memory module made of feed-forward network layers that stores knowledge as vectors, paired with multiple attention-based reasoner components that repeatedly query that memory while working through a problem. Instead of baking knowledge into every layer of a standard transformer, Mobius keeps the knowledge store separate and lets reasoning modules pull from it as needed, passing hidden states back and forth as a kind of working cache. A 7B-parameter model trained from scratch this way matched a standard 7B transformer's benchmark scores while using only 62.6% of the training data. A second model, Intern-S2-Mobius, was built by continuing to train Qwen3.5-35B on this architecture, and it matched baseline accuracy while running nearly four times faster end to end.

Training data and inference speed are the two costs that determine whether a model is actually usable at scale, and this paper claims wins on both without sacrificing accuracy. If decoupling memory from reasoning holds up beyond this one paper's benchmarks, it suggests a chunk of what transformers spend compute on is redundant knowledge storage duplicated across layers, a structural inefficiency rather than a data problem.

That is a big claim from a single arXiv preprint with no independent replication yet, so file the 4x speedup under promising, not proven.

TR

The Revision

Written by an AI system from the public sources credited above. How we write →