mlsanityData to 7 Oct 2026

LLM reasoning

17 directions · 17,737 papers · 99 this week

Busiest this week

Most new since Mon 5 Oct

Test-time compute and efficient thinking

2,144 papers18 this weekLow confidence in this grouping: the papers may belong to more than one topic, or the group may changeLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About

  1. Invariant-Measure Reasoners: Stable Representations for Latent Reasoning
  2. SemanticFold: Latent Sequence Compression SeparatesLanguage Modeling, Decodability, and Reasoning
  3. Unrolled Flow Models for Reasoning
  4. Learning Situation-Conditioned Thinking Policies for Long-Term LLM Agents
  5. Efficient Reasoning with Flow Language Models
Open →

All directions

Largest first · week from Mon 5 Oct

  1. Multimodal reasoningvisual, multimodal, reasoning, videoMedium confidence in this grouping: mostly one topic, with blurry edges · About2,6839Medium confidence in this grouping: mostly one topic, with blurry edges
  2. Reasoning benchmarks and evaluationlanguage, reasoning, large language, largeLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About2,3883Low confidence in this grouping: the papers may belong to more than one topic, or the group may change
  3. Test-time compute and efficient thinkingreasoning, test-time, efficient, scalingLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About2,14418Low confidence in this grouping: the papers may belong to more than one topic, or the group may change
  4. Reasoning for agents, planning and codereasoning, multi-agent, agents, llmLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About2,14314Low confidence in this grouping: the papers may belong to more than one topic, or the group may change
  5. Reinforcement learning for reasoningreinforcement, reinforcement learning, optimization, policyHigh confidence in this grouping: the papers form one clear topic · About1,65310High confidence in this grouping: the papers form one clear topic
  6. Spatial and embodied reasoningspatial, spatial reasoning, reasoning, embodiedMedium confidence in this grouping: mostly one topic, with blurry edges · About1,48311Medium confidence in this grouping: mostly one topic, with blurry edges
  7. Knowledge- and tool-augmented reasoningknowledge, graph, reasoning, retrievalMedium confidence in this grouping: mostly one topic, with blurry edges · About1,3053Medium confidence in this grouping: mostly one topic, with blurry edges
  8. Prompting and chain-of-thought reasoningchain-of-thought, distillation, reasoning, promptingLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About1,0027Low confidence in this grouping: the papers may belong to more than one topic, or the group may change
  9. Clinical and biomedical applicationsmedical, clinical, reasoning, medical reasoningMedium confidence in this grouping: mostly one topic, with blurry edges · About91710Medium confidence in this grouping: mostly one topic, with blurry edges
  10. Mathematical reasoningmathematical, mathematical reasoning, math, reasoningHigh confidence in this grouping: the papers form one clear topic · About7592High confidence in this grouping: the papers form one clear topic
  11. Audio-language model reasoningaudio, emotion, speech, audio-visualLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About3031Low confidence in this grouping: the papers may belong to more than one topic, or the group may change
  12. Chain-of-thought reasoning for drivingdriving, autonomous driving, autonomous, anomalyMedium confidence in this grouping: mostly one topic, with blurry edges · About2562Medium confidence in this grouping: mostly one topic, with blurry edges
  13. Causal reasoning in LLMscausal, causal reasoning, counterfactual, reasoningMedium confidence in this grouping: mostly one topic, with blurry edges · About1670Medium confidence in this grouping: mostly one topic, with blurry edges
  14. LLM reasoning for recommendationrecommendation, llm-based, recommender, generative recommendationHigh confidence in this grouping: the papers form one clear topic · About1600High confidence in this grouping: the papers form one clear topic
  15. Hallucination detection and mitigationhallucination, hallucinations, hallucination detection, detectionHigh confidence in this grouping: the papers form one clear topic · About1571High confidence in this grouping: the papers form one clear topic
  16. Diffusion reasoning and test-time scalingdiffusion, diffusion language, scaling, inference-time scalingHigh confidence in this grouping: the papers form one clear topic · About1291High confidence in this grouping: the papers form one clear topic
  17. Time series reasoningtime series, series, time, forecastingHigh confidence in this grouping: the papers form one clear topic · About880High confidence in this grouping: the papers form one clear topic
7 new this week aren’t in a direction yet
  1. Rethinking the Tradeoff Between Temporal Encoding and Nonlinear Computation in Spiking Language Models
  2. Where Can a Decision Model Diagnose HVAC Faults? Reasoning Demand, Physical Representation, and Robustness Under Shift
  3. Certified by Abstention: Distribution-Free Guarantees for Chain-of-Thought Verifiers at Small Calibration Budgets
  4. PEARS: Physical-Prior-Guided Efficient Adaptation via Failure Reasoning and Diffusion Steering for Tactile Manipulation
  5. OMIT the Action: Measuring Framing-Invariant Omission Bias under Philosophical Disagreement
  6. 2d-fet-bench: from spatial reasoning to fet design on flakes
  7. DeferKV: Rethinking Eviction Timing for One-Shot KV Cache Compression

Weekly activity

Papers per week, 22 Jun – 5 Oct (last week partial)

  1. DirectionConfidencepapers16 wks
  2. Reinforcement learning for reasoningHigh confidence in this grouping: the papers form one clear topic · AboutHigh confidence in this grouping: the papers form one clear topicPer week: 22 Jun: 14; 29 Jun: 13; 6 Jul: 8; 13 Jul: 12; 20 Jul: 17; 27 Jul: 13; 3 Aug: 18; 10 Aug: 11; 17 Aug: 13; 24 Aug: 16; 31 Aug: 16; 7 Sep: 10; 14 Sep: 10; 21 Sep: 24; 28 Sep: 55; 5 Oct: 101,653260
  3. Test-time compute and efficient thinkingLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · AboutLow confidence in this grouping: the papers may belong to more than one topic, or the group may changePer week: 22 Jun: 26; 29 Jun: 18; 6 Jul: 19; 13 Jul: 7; 20 Jul: 20; 27 Jul: 15; 3 Aug: 30; 10 Aug: 22; 17 Aug: 13; 24 Aug: 21; 31 Aug: 31; 7 Sep: 16; 14 Sep: 10; 21 Sep: 24; 28 Sep: 46; 5 Oct: 182,144336
  4. Reasoning for agents, planning and codeLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · AboutLow confidence in this grouping: the papers may belong to more than one topic, or the group may changePer week: 22 Jun: 20; 29 Jun: 10; 6 Jul: 20; 13 Jul: 12; 20 Jul: 18; 27 Jul: 24; 3 Aug: 17; 10 Aug: 19; 17 Aug: 12; 24 Aug: 20; 31 Aug: 24; 7 Sep: 10; 14 Sep: 27; 21 Sep: 22; 28 Sep: 35; 5 Oct: 142,143304
  5. Multimodal reasoningMedium confidence in this grouping: mostly one topic, with blurry edges · AboutMedium confidence in this grouping: mostly one topic, with blurry edgesPer week: 22 Jun: 33; 29 Jun: 29; 6 Jul: 23; 13 Jul: 21; 20 Jul: 13; 27 Jul: 30; 3 Aug: 30; 10 Aug: 23; 17 Aug: 24; 24 Aug: 26; 31 Aug: 28; 7 Sep: 17; 14 Sep: 12; 21 Sep: 23; 28 Sep: 49; 5 Oct: 92,683390
  6. Spatial and embodied reasoningMedium confidence in this grouping: mostly one topic, with blurry edges · AboutMedium confidence in this grouping: mostly one topic, with blurry edgesPer week: 22 Jun: 14; 29 Jun: 18; 6 Jul: 16; 13 Jul: 20; 20 Jul: 14; 27 Jul: 13; 3 Aug: 18; 10 Aug: 19; 17 Aug: 12; 24 Aug: 13; 31 Aug: 10; 7 Sep: 12; 14 Sep: 13; 21 Sep: 15; 28 Sep: 32; 5 Oct: 111,483250
  7. Prompting and chain-of-thought reasoningLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · AboutLow confidence in this grouping: the papers may belong to more than one topic, or the group may changePer week: 22 Jun: 7; 29 Jun: 7; 6 Jul: 11; 13 Jul: 8; 20 Jul: 6; 27 Jul: 4; 3 Aug: 17; 10 Aug: 4; 17 Aug: 4; 24 Aug: 12; 31 Aug: 10; 7 Sep: 5; 14 Sep: 7; 21 Sep: 11; 28 Sep: 19; 5 Oct: 71,002139
  8. Reasoning benchmarks and evaluationLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · AboutLow confidence in this grouping: the papers may belong to more than one topic, or the group may changePer week: 22 Jun: 16; 29 Jun: 7; 6 Jul: 9; 13 Jul: 21; 20 Jul: 20; 27 Jul: 9; 3 Aug: 18; 10 Aug: 12; 17 Aug: 11; 24 Aug: 18; 31 Aug: 13; 7 Sep: 12; 14 Sep: 7; 21 Sep: 10; 28 Sep: 23; 5 Oct: 32,388209
  9. Clinical and biomedical applicationsMedium confidence in this grouping: mostly one topic, with blurry edges · AboutMedium confidence in this grouping: mostly one topic, with blurry edgesPer week: 22 Jun: 12; 29 Jun: 13; 6 Jul: 3; 13 Jul: 6; 20 Jul: 9; 27 Jul: 9; 3 Aug: 6; 10 Aug: 10; 17 Aug: 9; 24 Aug: 8; 31 Aug: 16; 7 Sep: 8; 14 Sep: 4; 21 Sep: 9; 28 Sep: 20; 5 Oct: 10917152
  10. Knowledge- and tool-augmented reasoningMedium confidence in this grouping: mostly one topic, with blurry edges · AboutMedium confidence in this grouping: mostly one topic, with blurry edgesPer week: 22 Jun: 6; 29 Jun: 9; 6 Jul: 7; 13 Jul: 9; 20 Jul: 6; 27 Jul: 9; 3 Aug: 6; 10 Aug: 10; 17 Aug: 9; 24 Aug: 13; 31 Aug: 10; 7 Sep: 5; 14 Sep: 10; 21 Sep: 10; 28 Sep: 9; 5 Oct: 31,305131
  11. Audio-language model reasoningLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · AboutLow confidence in this grouping: the papers may belong to more than one topic, or the group may changePer week: 22 Jun: 2; 29 Jun: 4; 6 Jul: 3; 13 Jul: 1; 20 Jul: 5; 27 Jul: 1; 3 Aug: 2; 10 Aug: 2; 17 Aug: 1; 24 Aug: 5; 31 Aug: 0; 7 Sep: 2; 14 Sep: 6; 21 Sep: 5; 28 Sep: 10; 5 Oct: 130350
  12. Mathematical reasoningHigh confidence in this grouping: the papers form one clear topic · AboutHigh confidence in this grouping: the papers form one clear topicPer week: 22 Jun: 6; 29 Jun: 7; 6 Jul: 7; 13 Jul: 7; 20 Jul: 0; 27 Jul: 2; 3 Aug: 3; 10 Aug: 3; 17 Aug: 2; 24 Aug: 5; 31 Aug: 4; 7 Sep: 1; 14 Sep: 4; 21 Sep: 6; 28 Sep: 8; 5 Oct: 275967
  13. Diffusion reasoning and test-time scalingHigh confidence in this grouping: the papers form one clear topic · AboutHigh confidence in this grouping: the papers form one clear topicPer week: 22 Jun: 4; 29 Jun: 1; 6 Jul: 0; 13 Jul: 2; 20 Jul: 1; 27 Jul: 2; 3 Aug: 3; 10 Aug: 1; 17 Aug: 1; 24 Aug: 2; 31 Aug: 0; 7 Sep: 1; 14 Sep: 5; 21 Sep: 3; 28 Sep: 7; 5 Oct: 112934
  14. Chain-of-thought reasoning for drivingMedium confidence in this grouping: mostly one topic, with blurry edges · AboutMedium confidence in this grouping: mostly one topic, with blurry edgesPer week: 22 Jun: 5; 29 Jun: 5; 6 Jul: 2; 13 Jul: 3; 20 Jul: 1; 27 Jul: 4; 3 Aug: 5; 10 Aug: 6; 17 Aug: 4; 24 Aug: 3; 31 Aug: 5; 7 Sep: 2; 14 Sep: 1; 21 Sep: 4; 28 Sep: 0; 5 Oct: 225652
  15. Hallucination detection and mitigationHigh confidence in this grouping: the papers form one clear topic · AboutHigh confidence in this grouping: the papers form one clear topicPer week: 22 Jun: 2; 29 Jun: 2; 6 Jul: 3; 13 Jul: 0; 20 Jul: 2; 27 Jul: 2; 3 Aug: 1; 10 Aug: 2; 17 Aug: 0; 24 Aug: 0; 31 Aug: 0; 7 Sep: 0; 14 Sep: 1; 21 Sep: 2; 28 Sep: 3; 5 Oct: 115721
  16. Causal reasoning in LLMsMedium confidence in this grouping: mostly one topic, with blurry edges · AboutMedium confidence in this grouping: mostly one topic, with blurry edgesPer week: 22 Jun: 1; 29 Jun: 1; 6 Jul: 1; 13 Jul: 0; 20 Jul: 0; 27 Jul: 2; 3 Aug: 1; 10 Aug: 0; 17 Aug: 3; 24 Aug: 1; 31 Aug: 1; 7 Sep: 1; 14 Sep: 0; 21 Sep: 3; 28 Sep: 1; 5 Oct: 016716
  17. LLM reasoning for recommendationHigh confidence in this grouping: the papers form one clear topic · AboutHigh confidence in this grouping: the papers form one clear topicPer week: 22 Jun: 1; 29 Jun: 1; 6 Jul: 1; 13 Jul: 1; 20 Jul: 1; 27 Jul: 5; 3 Aug: 3; 10 Aug: 1; 17 Aug: 3; 24 Aug: 2; 31 Aug: 3; 7 Sep: 1; 14 Sep: 0; 21 Sep: 0; 28 Sep: 2; 5 Oct: 016025
  18. Time series reasoningHigh confidence in this grouping: the papers form one clear topic · AboutHigh confidence in this grouping: the papers form one clear topicPer week: 22 Jun: 0; 29 Jun: 0; 6 Jul: 3; 13 Jul: 0; 20 Jul: 0; 27 Jul: 1; 3 Aug: 3; 10 Aug: 1; 17 Aug: 0; 24 Aug: 1; 31 Aug: 0; 7 Sep: 1; 14 Sep: 0; 21 Sep: 0; 28 Sep: 2; 5 Oct: 08812