News

Notification update: Thanks to the large number of submissions, reviewing is taking longer than we planned. The notification of acceptance is postponed from July 31 to August 5, 2026 (AoE). All other dates remain unchanged. Thank you for your patience!
Deadline extension: By popular demand, the submission deadline has been extended by one week, from July 12 to July 19, 2026 (AoE). The notification of acceptance moves accordingly to July 31, 2026 (AoE).
Call for reviewers: We are actively looking for reviewers to join the program committee for the workshop. We encourage all interested researchers to apply, especially those from underrepresented groups. Prior reviewer experience is a nice-to-have, but not required. Interest and familiarity with subject matters related to the workshop is required. If you are interested, please fill out the application form to join us.
Call for papers: Submission portal open at OpenReview

About

Reasoning is rapidly moving beyond text-only benchmarks into multimodal, spatial, and embodied settings, where models must interpret long contexts, reason over visual scenes, plan actions, use tools, and interact with dynamic environments. This shift is especially visible in code agents, scientific assistants, embodied systems, and multimodal decision-making pipelines, where strong reasoning quality must now coexist with tight constraints on latency, memory, throughput, and serving cost. At the same time, many of the strongest gains in reasoning still come from more test-time compute, longer trajectories, or deeper search, which makes deployment substantially harder. The central question is therefore no longer only how to make models reason better, but how to make them reason efficiently, robustly, and at scale in the wild.

Building on the success of the 1st Workshop on Efficient Reasoning at NeurIPS 2025 (1,000+ attendees, 275 submissions), this second edition at COLM 2026 will bring together researchers and practitioners working across data, algorithms, systems, and real-world applications to tackle the challenges of making reasoning practical under real-world constraints.

Research Directions

Datasets & Evaluation

Curating data and benchmarks for long-context reasoning, symbolic planning, spatial reasoning, embodied reasoning, multimodal reasoning, and other challenging settings under resource constraints.

Algorithms

Efficient supervised fine-tuning, reinforcement learning, distillation, pruning, compression, progressive generation, and search-based reasoning that improve efficiency without sacrificing accuracy, faithfulness, or robustness.

Systems

Efficient RL training stacks, long-context and multimodal inference engines, KV-cache and memory management, quantized execution, and other systems advances that make reasoning practical on limited hardware.

Safety

Adversarial robustness of efficient reasoning pipelines, including backdoor attacks on compressed or distilled reasoning traces, and alignment degradation under pruning and quantization.

Applications

Deploying reasoning models in domains such as healthcare, embodied agents, autonomous driving, scientific discovery, and other safety-critical settings where latency and cost matter.

Call for Papers

The 2nd Workshop on Efficient Reasoning at COLM 2026 invites submissions from researchers working on making reasoning models faster, cheaper, and more efficient. We welcome contributions across data, algorithms, systems, safety, and applications, as well as interdisciplinary perspectives from the natural and social sciences.

Scope

We welcome contributions across a broad spectrum of topics, including but not limited to:

Pipelines for creating high-quality training data with resource constraints or deployment environments
Benchmark methodologies for assessing the efficiency and efficacy of reasoning models in real-world settings
Innovations in techniques for training reasoning models with a better trade-off between efficiency and performance
Approaches for accelerating inference through the design of algorithms and systems
Advancements in multimodal, spatial, and embodied reasoning under efficiency constraints
Theoretical analysis of the time and space complexity of reasoning models
Empirical investigations into the practical efficiency (latency, throughput, memory) of reasoning models
Efficient large-scale RL training systems and on-device inference engines
KV-cache compression and memory management for long-context reasoning
Safety and robustness of efficient reasoning pipelines
In-depth discussions exploring efficient deployments and applications of reasoning models

Submission Guidelines

Format: All submissions must be a single PDF file in the COLM submission format, with 4–10 pages of main text (references excluded). The main text must be self-contained. Reviewers are not required to read beyond the main text.

Submission Site: Submit papers through OpenReview.

Dual-submission and non-archival policy: We welcome ongoing and unpublished work. We will also accept papers that are under review at the time of submission, or that have been recently accepted, provided they do not breach any dual-submission or anonymity policies of those venues. The workshop is a non-archival venue and will not have official proceedings.

Double-blind reviewing: All submissions must be anonymized and may not contain any identifying information that may violate the double-blind reviewing policy. This policy applies to any supplementary or linked material as well, including code.

Contact: For any questions, please contact us at er.colm2026@gmail.com

Tentative Schedule

Full-day program with eight invited talks, five oral presentations, two poster sessions, and one panel discussion.

Morning Session

09:15 - 09:30	Opening Remarks
09:30 - 10:00	Invited Talk: Heng Ji (UIUC) — Building Scientific Worlds at Scale
10:00 - 10:15	Oral Presentation: Peiyang Xu — Context-Aware RL for Agentic and Multimodal LLMs
10:15 - 10:30	Oral Presentation: Nikhil Ranjan — Ignorance Is Shared, Knowledge Is Private
10:30 - 11:00	Invited Talk: Jiajun Wu (Stanford)
11:00 - 11:30	Invited Talk: Kai-Wei Chang (UCLA)
11:30 - 12:00	Invited Talk: Sewon Min (UC Berkeley)
12:00 - 13:30	Lunch Break & Poster Session 1

Afternoon Session

13:30 - 14:00	Invited Talk: Junjie Hu (University of Wisconsin) — Designing Memory for Test-Time Training
14:00 - 14:30	Invited Talk: Manling Li (Northwestern University)
14:30 - 14:45	Oral Presentation: Albert Gong — Express Language Modeling
14:45 - 15:00	Oral Presentation: Jeonghye Kim — Rebellious Student: Reversing Teacher Signals for Reasoning Exploration with Self-Distilled RLVR
15:00 - 15:15	Oral Presentation: Zheng Wang — PRESTO: Prefix-Aligned Tree Drafting for Diffusion Speculative Decoding
15:15 - 15:30	Afternoon Break
15:30 - 16:00	Panel Discussion with Yuandong Tian (Stealth AI, ex-Meta), Kaitao Song (Apple), and Ying Liu (Scale AI)
16:00 - 16:30	Invited Talk: Niklas Muennighoff (Stanford) — Infinite Test-Time Scaling
16:30 - 17:00	Invited Talk: Rulin Shao (University of Washington) — Context Language Models
17:00 - 18:00	Poster Session 2

Submission Deadline	~~July 12, 2026~~ July 19, 2026 (AoE)
Notification of Acceptance	~~July 24, 2026~~ ~~July 31, 2026~~ August 5, 2026 (AoE)
Camera-ready Deadline	TBD
Workshop Day	October 9, 2026
Location	Hilton Union Square, San Francisco

News

About

Research Directions

Call for Papers

Scope

Submission Guidelines

Key Dates

Tentative Schedule

Morning Session

Afternoon Session

Confirmed Invited Speakers

Kai-Wei Chang

Junjie Hu

Heng Ji

Sewon Min

Niklas Muennighoff

Yuandong Tian

Kaitao Song

Ying Liu

Jiajun Wu

Manling Li

Rulin Shao

Workshop Organizers

Zefan Cai

Cheng Luo

Yijiang Li

Weijia Shi

Songlin Yang

Jiawei Zhao

Zhen Dong

Haozheng Luo

Zhijian Liu

Beidi Chen

Jiahao Yu

Rui-jie Zhu

Xinnan Dai

Shenglai Zeng

Yiping Lu

Sponsors

Previous Editions

1st Workshop on Efficient Reasoning @ NeurIPS 2025

1st Workshop on Efficient Spatial Reasoning @ ICLR 2026