We are pleased to announce the Latent & Implicit Thinking – Going Beyond CoT Reasoning (LIT) workshop at ICLR 2026. While chain-of-thought (CoT) reasoning aligns intuitively with human cognition, it may not represent the most effective internal computation medium for neural networks. Emerging research is exploring alternative forms of reasoning, including continuous hidden representations, discrete reasoning tokens optimized for efficiency rather than readability, and non-autoregressive approaches such as diffusion-based thinking. These approaches open up exciting possibilities for more efficient and potentially more capable implicit reasoning in future AI systems. The LIT workshop aims to bring together researchers across the spectrum of implicit reasoning to spark collaboration and accelerate progress in this emerging area.
Submit Your Paper
| Submit via OpenReview → | Call for Papers → | Call for Reviewers → |
Submission deadline: February 5, 2026 (AOE) extended to Feb 8 AOE, but strongly encourage to submit before Feb 5 AOE
We welcome submissions on topics including training strategies for implicit reasoning, looped architectures, mechanical interpretability, special thinking tokens, KV-cache and hybrid stateful reasoning, inference paradigms beyond autoregression, theoretical results, evaluation and benchmarks, and limitations and safety. See the Call for Papers for a complete list of topics.
Important Dates
- Submission deadline:
February 5, 2026 (AOE)extended to Feb 8 AOE, but strongly encourage to submit before Feb 5 AOE - Reviewing period: February 8 - February 27, 2026
- Notification of acceptance: March 1, 2026, 11:59pm AoE
- Camera-ready deadline: March 10, 2026
- Workshop date: April 26-27, 2026 (Rio de Janeiro, following the ICLR 2026 main conference)