{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:W5HOEO55XSCA5OT2L5X5JMLG7D","short_pith_number":"pith:W5HOEO55","schema_version":"1.0","canonical_sha256":"b74ee23bbdbc840eba7a5f6fd4b166f8edfc5df31c739b840f04c5df7b535c9b","source":{"kind":"arxiv","id":"2410.10629","version":3},"attestation_state":"computed","paper":{"title":"SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformers","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Sana-0.6B generates high-resolution images competitively with 12B-parameter models while running over 100 times faster on consumer GPUs.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Enze Xie, Han Cai, Haotian Tang, Junsong Chen, Junyu Chen, Ligeng Zhu, Muyang Li, Song Han, Yao Lu, Yujun Lin, Zhekai Zhang","submitted_at":"2024-10-14T15:36:42Z","abstract_excerpt":"We introduce Sana, a text-to-image framework that can efficiently generate images up to 4096$\\times$4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU. Core designs include: (1) Deep compression autoencoder: unlike traditional AEs, which compress images only 8$\\times$, we trained an AE that can compress images 32$\\times$, effectively reducing the number of latent tokens. (2) Linear DiT: we replace all vanilla attention in DiT with linear attention, which is more efficient at high resolu"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2410.10629","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2024-10-14T15:36:42Z","cross_cats_sorted":[],"title_canon_sha256":"e95928638dc8b30e52d7551e409e10ea8ac631fff31b5419ab2807f3683bd070","abstract_canon_sha256":"15afb39d277185e4366dd1119febf88c6db3f739196fb284b0c253610968d58f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:05.190527Z","signature_b64":"avjaSPFXO2OZr70FL6fha0OdL8nCdDjGdpkwT/3uCI82FDS9+SqS3y6mDIfPJDxWaTu1Tpz+x7bP/XzrZ9k8Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b74ee23bbdbc840eba7a5f6fd4b166f8edfc5df31c739b840f04c5df7b535c9b","last_reissued_at":"2026-05-17T23:39:05.189950Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:05.189950Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformers","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Sana-0.6B generates high-resolution images competitively with 12B-parameter models while running over 100 times faster on consumer GPUs.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Enze Xie, Han Cai, Haotian Tang, Junsong Chen, Junyu Chen, Ligeng Zhu, Muyang Li, Song Han, Yao Lu, Yujun Lin, Zhekai Zhang","submitted_at":"2024-10-14T15:36:42Z","abstract_excerpt":"We introduce Sana, a text-to-image framework that can efficiently generate images up to 4096$\\times$4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU. Core designs include: (1) Deep compression autoencoder: unlike traditional AEs, which compress images only 8$\\times$, we trained an AE that can compress images 32$\\times$, effectively reducing the number of latent tokens. (2) Linear DiT: we replace all vanilla attention in DiT with linear attention, which is more efficient at high resolu"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Sana-0.6B is very competitive with modern giant diffusion model (e.g. Flux-12B), being 20 times smaller and 100+ times faster in measured throughput.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The 32x deep-compression autoencoder preserves sufficient perceptual quality and text alignment at high resolutions without introducing artifacts that linear attention cannot correct.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Sana-0.6B produces high-resolution images with strong text alignment at 20x smaller size and 100x higher throughput than Flux-12B by combining 32x image compression, linear DiT blocks, and a decoder-only LLM text encoder.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Sana-0.6B generates high-resolution images competitively with 12B-parameter models while running over 100 times faster on consumer GPUs.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"b59caa39f172598343a38d0cc619ff6837b0e3aaf67b4883cf3057076ac0fbaa"},"source":{"id":"2410.10629","kind":"arxiv","version":3},"verdict":{"id":"dc9f5c27-201c-4f60-b8ef-12c3f83aef97","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T00:52:14.386983Z","strongest_claim":"Sana-0.6B is very competitive with modern giant diffusion model (e.g. Flux-12B), being 20 times smaller and 100+ times faster in measured throughput.","one_line_summary":"Sana-0.6B produces high-resolution images with strong text alignment at 20x smaller size and 100x higher throughput than Flux-12B by combining 32x image compression, linear DiT blocks, and a decoder-only LLM text encoder.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The 32x deep-compression autoencoder preserves sufficient perceptual quality and text alignment at high resolutions without introducing artifacts that linear attention cannot correct.","pith_extraction_headline":"Sana-0.6B generates high-resolution images competitively with 12B-parameter models while running over 100 times faster on consumer GPUs."},"references":{"count":29,"sample":[{"doi":"","year":null,"title":"eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers","work_id":"2cd7b629-ab37-4ce5-b51e-aa4d99547468","ref_index":1,"cited_arxiv_id":"2211.01324","is_internal_anchor":true},{"doi":"","year":2022,"title":"All are worth words: a vit backbone for score-based diffusion models","work_id":"98f35013-1df4-4a37-8b71-6206599f9833","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2005,"title":"Language Models are Few-Shot Learners","work_id":"214732c0-2edd-44a0-af9e-28184a2b8279","ref_index":3,"cited_arxiv_id":"2005.14165","is_internal_anchor":true},{"doi":"","year":2020,"title":"Mahoney, and Kurt Keutzer","work_id":"911a6b35-56a0-42b3-9951-b6f6ac9da2ed","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"PixArt- : Weak-to-strong training of diffusion transformer for 4k text-to-image generation","work_id":"5dc0cfb7-2014-4631-be70-f632bf6ca9ec","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":29,"snapshot_sha256":"8dab2fbac73eaa14a78aa7dcdf69c72b55b8576ffa2200f5345add58a55a44ee","internal_anchors":11},"formal_canon":{"evidence_count":1,"snapshot_sha256":"a7925ba5afbca173b270de0c9e79d4db4aa675988ba6f9d8760e372d6fde38d2"},"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":"2410.10629","created_at":"2026-05-17T23:39:05.190071+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.10629v3","created_at":"2026-05-17T23:39:05.190071+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.10629","created_at":"2026-05-17T23:39:05.190071+00:00"},{"alias_kind":"pith_short_12","alias_value":"W5HOEO55XSCA","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"W5HOEO55XSCA5OT2","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"W5HOEO55","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":36,"internal_anchor_count":36,"sample":[{"citing_arxiv_id":"2411.16748","citing_title":"Multimodal Diffusion Transformer with Memory Bank for Scalable Long-Duration Talking Video Generation","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22777","citing_title":"DecQ: Detail-Condensing Queries for Enhanced Reconstruction and Generation in Representation Autoencoders","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2603.21002","citing_title":"SURF: Signature-Retained Fast Video Generation","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2602.06886","citing_title":"Prompt Reinjection: Alleviating Prompt Forgetting in Multimodal Diffusion Transformers","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20808","citing_title":"Spatial Gram Alignment for Ultra-High-Resolution Image Synthesis","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20708","citing_title":"Rethinking Cross-Layer Information Routing in Diffusion Transformers","ref_index":59,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21272","citing_title":"MONET: A Massive, Open, Non-redundant and Enriched Text-to-image dataset","ref_index":102,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05204","citing_title":"D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models","ref_index":105,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15682","citing_title":"DreamSR: Towards Ultra-High-Resolution Image Super-Resolution via a Receptive-Field Enhanced Diffusion Transformer","ref_index":63,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20147","citing_title":"PixVerve: Advancing Native UHR Image Generation to 100MP with a Large-Scale High-Quality Dataset","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2506.23690","citing_title":"SynMotion: Semantic-Visual Adaptation for Motion Customized Video Generation","ref_index":93,"is_internal_anchor":true},{"citing_arxiv_id":"2509.22126","citing_title":"Guidance Watermarking for Diffusion Models","ref_index":43,"is_internal_anchor":true},{"citing_arxiv_id":"2604.09568","citing_title":"EvoDiagram: Agentic Editable Diagram Creation via Design Expertise Evolution","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15178","citing_title":"SANA-WM: Efficient Minute-Scale World Modeling with Hybrid Linear Diffusion Transformer","ref_index":82,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14569","citing_title":"Bridging Brain and Semantics: A Hierarchical Framework for Semantically Enhanced fMRI-to-Video Reconstruction","ref_index":112,"is_internal_anchor":true},{"citing_arxiv_id":"2604.08586","citing_title":"FluidFlow: a flow-matching generative model for fluid dynamics surrogates on unstructured meshes","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13047","citing_title":"Revealing the Gap in Human and VLM Scene Perception through Counterfactual Semantic Saliency","ref_index":62,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06061","citing_title":"PromptEvolver: Prompt Inversion through Evolutionary Optimization in Natural-Language Space","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12013","citing_title":"L2P: Unlocking Latent Potential for Pixel Generation","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2601.03233","citing_title":"LTX-2: Efficient Joint Audio-Visual Foundation Model","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09313","citing_title":"Attention Sinks in Diffusion Transformers: A Causal Analysis","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09313","citing_title":"Attention Sinks in Diffusion Transformers: A Causal Analysis","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10019","citing_title":"The two clocks and the innovation window: When and how generative models learn rules","ref_index":80,"is_internal_anchor":true},{"citing_arxiv_id":"2604.24351","citing_title":"Diffusion Templates: A Unified Plugin Framework for Controllable Diffusion","ref_index":44,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05646","citing_title":"MUSE: Resolving Manifold Misalignment in Visual Tokenization via Topological Orthogonality","ref_index":148,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":1,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D","json":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D.json","graph_json":"https://pith.science/api/pith-number/W5HOEO55XSCA5OT2L5X5JMLG7D/graph.json","events_json":"https://pith.science/api/pith-number/W5HOEO55XSCA5OT2L5X5JMLG7D/events.json","paper":"https://pith.science/paper/W5HOEO55"},"agent_actions":{"view_html":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D","download_json":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D.json","view_paper":"https://pith.science/paper/W5HOEO55","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.10629&json=true","fetch_graph":"https://pith.science/api/pith-number/W5HOEO55XSCA5OT2L5X5JMLG7D/graph.json","fetch_events":"https://pith.science/api/pith-number/W5HOEO55XSCA5OT2L5X5JMLG7D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D/action/storage_attestation","attest_author":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D/action/author_attestation","sign_citation":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D/action/citation_signature","submit_replication":"https://pith.science/pith/W5HOEO55XSCA5OT2L5X5JMLG7D/action/replication_record"}},"created_at":"2026-05-17T23:39:05.190071+00:00","updated_at":"2026-05-17T23:39:05.190071+00:00"}