mlsanityData to 7 Oct 2026

Test-time Recursive Thinking: Self-Improvement without External Feedback

arXiv 2602.03094 ↗First posted 3 Feb 2026cs.CLcited by 7

On the map: LLM reasoning → Reinforcement learning for reasoning

also in 1 more
Abstract (arXiv)

Modern Large Language Models (LLMs) have shown rapid improvements in reasoning capabilities, driven largely by reinforcement learning (RL) with verifiable rewards. Here, we ask whether these LLMs can self-improve without the need for additional training. We identify two core challenges for such systems: (i) efficiently generating diverse, high-quality candidate solutions, and (ii) reliably selecting correct answers in the absence of ground-truth supervision. To address these challenges, we propose Test-time Recursive Thinking (TRT), an iterative self-improvement framework that conditions generation on rollout-specific strategies, accumulated knowledge, and self-generated verification signals. Using TRT, open-source models reach 100% accuracy on AIME-25/24, and on LiveCodeBench's most difficult problems, closed-source models improve by 10.4-14.8 percentage points without external feedback.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning. Builds on 25 of its 35 references that are in the map. First posted on arXiv 3 Feb 2026.

Builds on

25 of 35 references.

Closest ancestors · 4

Same direction, newest first

  1. 2025ExGRPO: Learning to Reason from Experience2 Oct
  2. 2025GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning25 Jul
  3. 2025Optimizing Test-Time Compute via Meta Reinforcement Fine-Tuning10 Mar
  4. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning22 Jan

Other references · 21

  1. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  3. 2025Qwen3 Technical Reportcited by 8,195
  4. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  5. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
  6. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,928
  7. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 1,906
  8. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,258
  9. 2024OpenAI o1 System Cardcited by 1,220
  10. 2016RL2: Fast Reinforcement Learning via Slow Reinforcement Learningcited by 490
  11. 2016Learning to reinforcement learncited by 369
  12. 2025Agentic Context Engineering: Evolving Contexts for Self-Improving Language Modelscited by 349
9 more
  1. 2025ThetaEvolve: Test-time Learning on Open Problemscited by 52
  2. 2025Recursive Self-Aggregation Unlocks Deep Thinking in Large Language Modelscited by 30
  3. 2025RLAD: Training LLMs to Discover Abstractions for Solving Reasoning Problemscited by 21
  4. 2025Meta-RL Induces Exploration in Language Agentscited by 17
  5. 2025The Markovian Thinker: Architecture-Agnostic Linear Scaling of Reasoningcited by 12
  6. 2025The Sequential Edge: Inverse-Entropy Voting Beats Parallel Self-Consistency at Matched Computecited by 11
  7. 2026Human-AI Co-design for Clinical Prediction Modelscited by 9
  8. 2025Scalable Oversight for Superhuman AI via Recursive Self-Critiquingcited by 4
  9. 2025MatryoshkaThinking: Recursive Test-Time Scaling Enables Efficient Reasoningcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 7.

7 papers.

  1. 2026Scaling Test-Time Compute for Agentic Codingcited by 11
  2. 2026Federation over Text: Insight Sharing for Multi-Agent Reasoningcited by 2
  3. 2026Test-Time Learning with an Evolving Librarycited by 2
  4. 2026Test-time Scaling over Perception: Resolving the Grounding Paradox in Thinking with Imagescited by 1
  5. 2026Agentic-imodels: Evolving agentic interpretability tools via autoresearchcited by 1
  6. 2026Self-Verified Distillation: Your Language Model Is Secretly Its Own Synthetic Data Pipelinecited by 1
  7. 2026Register Tokens for Bounded-State Reasoning in Diffusion Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Can Large Language Models Invent Algorithms to Improve Themselves?: Algorithm Discovery for Recursive Self-Improvement through Reinforcement Learning
  2. 2026Re2: Unlocking LLM Reasoning via Reinforcement Learning with Re-solving
  3. 2026TTCS: Test-Time Curriculum Synthesis for Self-Evolving
  4. 2026A Task-Centric Theory for Iterative Self-Improvement with Easy-to-Hard Curricula
  5. 2025Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning
  6. 2025Incentivizing LLMs to Self-Verify Their Answers
  7. 2026TTSR: Test-Time Self-Evolving via Reflection
  8. 2025Rethinking Thinking Tokens: LLMs as Improvement Operators