Skip to demo

Constant memory.
Unbounded context.

AwareLiquid is a bio-inspired cognitive AI framework. The attention-free O-Series carries a fixed-size state instead of a growing KV-cache, so context length stops dictating memory.

Carried state0.381 MB
Context1M tokens
vs KV-cache8063× smaller
M-Series M1 Cognitive · GWT + Sleep · Slow Thinking
O-Series O1 Edge · Liquid ODE · Fast Thinking
8063× Less inference memory at 1M tokens
O-series state vs. KV-cache
0.381 MB Constant carried state
flat from 512 to 1M tokens
0.56 / 0.00 Cross-window recall
fast-weight vs. attention/LoRA
48M From-scratch edge model
attention-free O-series
Bio-inspired Architecture

Three Mechanisms That Make Thinking Possible

AwareLiquid stacks three neuroscience-derived subsystems on a continuous-time liquid core — each solving a distinct cognitive problem that Transformers and SSMs cannot address.

MT-LNN architecture pipeline: embedding → LTC core → liquid core modules (GWT, Sleep, Hebbian, Predictive Coding) → output → EWC continual learning
AwareLiquid M1 end-to-end pipeline — continuous-time LTC core + GWT competitive broadcast + sleep consolidation + EWC online learning
M1 · M-Series

Global Workspace Theory (GWT)

Specialist cortical modules compete for a shared broadcast channel — the "global workspace." The winner propagates its representation to all other modules, enabling selective attention, emergent integration, and conscious-like information routing.

M1 · M-Series

Sleep Consolidation (NREM Bridge)

Synthetic NREM cycles replay short-term session activations and compress them into a persistent graph-memory structure. Synaptic Homeostasis (SHY) downscaling prunes redundant connections, boosting signal-to-noise. The model grows smarter during rest.

O1 · O-Series

Liquid ODE Dynamics

Hidden state evolves as a continuous-time differential equation with learnable time constants — inspired by microtubule dynamics. No discrete layer stacking, no KV-cache growth. Constant-memory state enables edge deployment at millisecond latency.

Engineering Reality

Measured.
Reproducible.

Every claim below is measured on real runs and backed by open code and data.

CapabilityResult
O(1) inference memory (O-series)0.381 MB flat → 8063× smaller @1M tokens
Cross-window associative recall0.56 vs 0.000 (attention/LoRA)
Cross-session snapshot → restorebit-exact lossless
Stable training at 125M scaleno NaN, 20K steps, 3 seeds
Irregular-sampling robustness (battery SoH)4× steadier than LSTM/GRU when samples drop

Full methodology, raw data and reproduction scripts: BENCHMARKS.md.

Full methodology & results →
Line chart: Transformer KV-cache grows linearly O(n) with context length while the recurrent O-series state stays flat — constant memory regardless of context
Inference memory vs. context length. The KV-cache grows O(n); the O-series carried state is flat. Measured to 1M tokens: 0.381 MB state vs. 3,072 MB KV-cache — 8063× smaller. This measures inference carried-state bytes only — it is not evidence of model quality beyond the 512-token training length (out-of-window LM results are null; see RESULTS.md).
Live Demo

Try Both Models in Your Browser

Two distinct models, both live. M1 (large, cognitive) runs trained liquid-core residual adapters on a frozen TinyLlama-1.1B-Chat base — a general assistant. O1 (small, edge) is the native 48M liquid architecture trained from scratch with no base model — the raw backbone for hardware / always-on inference. Switch between them below. Ask in English or Chinese.

Connecting…

Connecting…

Research-grade demo — quality is experimental. What to expect →

Open Full-Page Demo →

The Multi-Agent Future

A Native Substrate for Cognitive Agents

Transformers excel at static retrieval. AwareLiquid is built for something harder: agents that form episodic memories across sessions, coordinate through shared global workspace broadcasts, and continuously consolidate knowledge without catastrophic forgetting. The foundation for an Awareness Market of collaborating cognitive agents.

See current architecture →
Positioning

Same Lineage. Different Frontier.

Both projects build on MIT liquid neural networks. Liquid AI solved static-weight inference efficiency. AwareLiquid asks the next question: can a model acquire new knowledge without erasing the old?

Liquid AI · LFM2
Maximum Inference
Efficiency
  • SSM + linear attention + MoE hybrid
  • 350M – 24B parameters, commercial scale
  • Weights frozen post-training; in-context only
  • Targets industrial time-series & edge devices
vs
AwareLiquid · M1 / O1
Cognitive Continuity
& Lifelong Learning
  • Continuous-time RNN + GWT + Sleep + EWC
  • 48M – 60.7M, open research prototype
  • Online weight updates — learns without forgetting
  • Targets episodic memory, agent cognition & adaptation