Loading patterns…
Latent Recurrent Thinking(LRT)
Iterative reasoning within continuous latent space without explicit token generation
In 30 seconds
- What
- Reasons iteratively in continuous latent space by unrolling a shared recurrent block, skipping intermediate token generation until final decoding.
- When to use
- Complex problems requiring deep exploration of solution space where intermediate reasoning steps don't need human inspection or external feedback.
- Watch out
- Latent reasoning is opaque; you cannot inspect, debug, or steer intermediate steps if the final answer goes wrong.
Ask the AI expert about this pattern
Opens the assistant with your question prefilled. You review it before sending.
Latent Recurrent Thinking: Overview
Iterative reasoning within continuous latent space without explicit token generation
- Reasons in continuous latent space instead of emitting CoT tokens
- Dynamic reasoning depth adjustment
- Parallel pathway exploration in latent space
- No intermediate token generation required
- Scales test-time compute by unrolling a shared recurrent block
- Mimics human intuitive thinking processes
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.
References
The papers, specifications, and repositories this pattern is based on.
- Geiping et al. (2025): Scaling up Test-Time Compute with Latent Reasoning (Recurrent Depth)
From the engineer behind this catalog
Get your agent architecture reviewed
This page documents one pattern. Your system runs dozens, and most failures live in how they fit together. Have the whole design reviewed against the 288 patterns in this catalog: architecture, reliability, evaluation and cost, every finding mapped to the pattern that fixes it.
€750 instead of €1,500, one week, written report and walkthrough call, until 30 September