{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IM4WIIM2SG3QPKR43Z7SC23A6N","short_pith_number":"pith:IM4WIIM2","schema_version":"1.0","canonical_sha256":"433964219a91b707aa3cde7f216b60f378915f392866daf70e51364245161c28","source":{"kind":"arxiv","id":"2504.09858","version":1},"attestation_state":"computed","paper":{"title":"Reasoning Models Can Be Effective Without Thinking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Charlie Snell, Jingxuan He, Matei Zaharia, Sewon Min, Tyler Griggs, Wenjie Ma","submitted_at":"2025-04-14T04:08:16Z","abstract_excerpt":"Recent LLMs have significantly improved reasoning capabilities, primarily by including an explicit, lengthy Thinking process as part of generation. In this paper, we question whether this explicit thinking is necessary. Using the state-of-the-art DeepSeek-R1-Distill-Qwen, we find that bypassing the thinking process via simple prompting, denoted as NoThinking, can be surprisingly effective. When controlling for the number of tokens, NoThinking outperforms Thinking across a diverse set of seven challenging reasoning datasets--including mathematical problem solving, formal theorem proving, and co"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2504.09858","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-04-14T04:08:16Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"72c7492c0782af475d5662659d57aa10faa54527f9088875d0b5b8432a0ef924","abstract_canon_sha256":"6578d6c2264c91036961926bb9218e9538fcc0a287e81228a02e4dc1db6408ce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:48:43.873727Z","signature_b64":"RF5mmeodqwHIfqNlejk3AU9ieoIN/KLHg3KTB2s46lnJCGUOusQttOj13JUlAKj/G4DaSi8e9gV16TIjhm29AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"433964219a91b707aa3cde7f216b60f378915f392866daf70e51364245161c28","last_reissued_at":"2026-07-05T10:48:43.873264Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:48:43.873264Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reasoning Models Can Be Effective Without Thinking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Charlie Snell, Jingxuan He, Matei Zaharia, Sewon Min, Tyler Griggs, Wenjie Ma","submitted_at":"2025-04-14T04:08:16Z","abstract_excerpt":"Recent LLMs have significantly improved reasoning capabilities, primarily by including an explicit, lengthy Thinking process as part of generation. In this paper, we question whether this explicit thinking is necessary. Using the state-of-the-art DeepSeek-R1-Distill-Qwen, we find that bypassing the thinking process via simple prompting, denoted as NoThinking, can be surprisingly effective. When controlling for the number of tokens, NoThinking outperforms Thinking across a diverse set of seven challenging reasoning datasets--including mathematical problem solving, formal theorem proving, and co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.09858","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2504.09858/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2504.09858","created_at":"2026-07-05T10:48:43.873320+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.09858v1","created_at":"2026-07-05T10:48:43.873320+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.09858","created_at":"2026-07-05T10:48:43.873320+00:00"},{"alias_kind":"pith_short_12","alias_value":"IM4WIIM2SG3Q","created_at":"2026-07-05T10:48:43.873320+00:00"},{"alias_kind":"pith_short_16","alias_value":"IM4WIIM2SG3QPKR4","created_at":"2026-07-05T10:48:43.873320+00:00"},{"alias_kind":"pith_short_8","alias_value":"IM4WIIM2","created_at":"2026-07-05T10:48:43.873320+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":22,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00053","citing_title":"SWE-Router: Routing in Multi-turn Agentic Software Engineering Tasks","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07108","citing_title":"DyCon: Dynamic Reasoning Control via Evolving Difficulty Modeling","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03503","citing_title":"ThoughtFold: Folding Reasoning Chains via Introspective Preference Learning","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02835","citing_title":"Thinking Past the Answer: Evaluating Harmful Overthinking in Large Reasoning Models","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14358","citing_title":"Uncovering the Representation Geometry of Minimal Cores in Overcomplete Reasoning Traces","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17890","citing_title":"Dynamic Rollout Editing for Reducing Overthinking in RL-Trained Reasoning Models","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17672","citing_title":"Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19358","citing_title":"Taming the Thinker: Conditional Entropy Shaping for Adaptive LLM Reasoning","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19316","citing_title":"A Multi-Agent Framework for Feature-Constrained Difficulty Control in Reading Comprehension Item Generation","ref_index":103,"is_internal_anchor":false},{"citing_arxiv_id":"2509.05489","citing_title":"Self-Aligned Reward: Towards Effective and Efficient Reasoners","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2509.25758","citing_title":"Thinking Sparks!: Emergent Attention Heads in Reasoning Models During Post Training","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2510.24941","citing_title":"Can Aha Moments Be Fake? Towards Quantifying Decorative and True Thinking in Chain-of-Thought","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2506.06941","citing_title":"The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2506.13757","citing_title":"AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2503.16419","citing_title":"Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models","ref_index":129,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10564","citing_title":"DeepSight: Long-Horizon World Modeling via Latent States Prediction for End-to-End Autonomous Driving","ref_index":105,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06165","citing_title":"Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost","ref_index":198,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14142","citing_title":"From $P(y|x)$ to $P(y)$: Investigating Reinforcement Learning in Pre-train Space","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07686","citing_title":"The Coupling Tax: How Shared Token Budgets Undermine Visible Chain-of-Thought Under Fixed Output Limits","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16890","citing_title":"Step-GRPO: Internalizing Dynamic Early Exit for Efficient Reasoning","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17304","citing_title":"Efficient Test-Time Scaling via Temporal Reasoning Aggregation","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21027","citing_title":"HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering","ref_index":66,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N","json":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N.json","graph_json":"https://pith.science/api/pith-number/IM4WIIM2SG3QPKR43Z7SC23A6N/graph.json","events_json":"https://pith.science/api/pith-number/IM4WIIM2SG3QPKR43Z7SC23A6N/events.json","paper":"https://pith.science/paper/IM4WIIM2"},"agent_actions":{"view_html":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N","download_json":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N.json","view_paper":"https://pith.science/paper/IM4WIIM2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.09858&json=true","fetch_graph":"https://pith.science/api/pith-number/IM4WIIM2SG3QPKR43Z7SC23A6N/graph.json","fetch_events":"https://pith.science/api/pith-number/IM4WIIM2SG3QPKR43Z7SC23A6N/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N/action/storage_attestation","attest_author":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N/action/author_attestation","sign_citation":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N/action/citation_signature","submit_replication":"https://pith.science/pith/IM4WIIM2SG3QPKR43Z7SC23A6N/action/replication_record"}},"created_at":"2026-07-05T10:48:43.873320+00:00","updated_at":"2026-07-05T10:48:43.873320+00:00"}