Research
Subconscious is an AI lab making open models far more capable.
We build TIMRUN, an inference runtime designed for agentic work, and the complementary post-trained TIM family of models.
Our thesis
Agents are bottlenecked by the context window.
The problem
Models get expensive, slow, and unreliable as the context window fills up.
Agents lose track of earlier reasoning as their context window fills, and waste compute reprocessing tokens they have already seen. General-purpose inference systems make it worse: every compaction invalidates the prefix cache, exactly when traces are longest.
Our approach
Solve the context problem at the runtime and model layer on the GPU, not the app layer.
We built an inference runtime that treats agent traces as the primary workload, and post-train open models to reason and use tools on it well. The two are co-designed to keep an agent's working context compact and relevant, so the system reasons further, faster, on the same hardware.
Pillar 1: Inference Runtime
An inference system designed for agents
General-purpose inference systems waste compute on agent workloads. We built TIMRUN, a runtime that treats agent traces as the primary workload for better performance and reliability.
Featured · June 2026
Subconscious Cache
Our system for KV cache pruning with preserved suffix states. When a span of context is pruned, the information it contributed survives in the retained suffix states instead of being thrown away and re-encoded.
50%
Lower latency per task
2.3×
Supported concurrent workloads
3.5×
Faster token generation
Auto Compaction
Our runtime enables models to prune their own message list at inference time. Compaction happens automatically inside the familiar chat completions and messages APIs, and the Subconscious Cache preserves memory across those prunes.
Read the postTIMRUN in Production
This is more than research, it's live in production today. Customers utilize our inference runtime via API, on their own GPU clusters, and on local devices.
Read the postSTANDARD RUNTIME · PREFIX CACHE ONLY
Everything after the pruned span is re-encoded from scratch.
SUBCONSCIOUS TIMRUN · PREFIX + SUFFIX CACHE
The suffix is reused too. Some of the pruned token information is preserved in the suffix.
Each strip is one agent context after a mid-conversation prune. A standard prefix cache only reuses tokens before the pruned span, so everything after it is re-encoded. TIMRUN also reuses the suffix, keeping most of the context cached.
Pillar 2: Model Post-training
Redline
TIMRUN alone is powerful. Redline is how open models learn to use it even better.
01
Co-designed with the runtime
Redline uses TIMRUN under the hood and trains the models to exploit context compression and suffix reuse.
02
Highly efficient post-training
With context compression during training, Redline requires up to 75% less compute to train on huge reasoning trajectories. That cost saving flows straight through to our customers.
03
SFT, OPD, and RL on agentic workloads
We use it to run SFT, OPD, and RL on open-weight models on long-horizon, tool-heavy tasks. Coding agents, browser control, workflow automation, research tasks, and more.
04
For our TIM models, or yours
Redline is how we create our TIM family of models, but also how we post-train models for customers on their unique data and tooling.
Born at MIT CSAIL
Publications
The work above stands on research we published in the MIT ecosystem.
Open collaboration
We publish, and we partner.
Our research is informed by our connections to the academic community and our commitment to publishing our findings. Today we collaborate with labs at MIT, Princeton, Harvard, Carnegie Mellon, Tel Aviv, and Berkeley.
For research collaborations, academic partnerships, or questions about our work, reach out to hongyin [at] [this website domain].