Spaces:
Running
Running
| { | |
| "meta": { | |
| "title": "StreamAV-Bench Leaderboard", | |
| "tagline": "Streaming systems, evaluated while they run.", | |
| "description": "Official metric-wise results for streaming audio-video generation across progressive and interactive evaluation tracks.", | |
| "systemCount": 13, | |
| "metricCount": 32, | |
| "scenarioCount": 320, | |
| "durationSeconds": 180, | |
| "publishedPrecision": 3, | |
| "links": { | |
| "project": "https://liukqchoco.github.io/StreamAVBench/", | |
| "code": "https://github.com/liukqchoco/StreamAVBench-Code", | |
| "paper": "https://arxiv.org/abs/2608.26336", | |
| "dataset": "https://huggingface.co/datasets/StreamAVBench/StreamAVBench" | |
| } | |
| }, | |
| "tracks": [ | |
| { | |
| "id": "shared", | |
| "label": "Shared", | |
| "count": 9, | |
| "description": "Quality, alignment, continuity, and efficiency metrics applied across both tracks." | |
| }, | |
| { | |
| "id": "progressive", | |
| "label": "Progressive", | |
| "count": 12, | |
| "description": "Instruction adherence and long-horizon stability under one global prompt." | |
| }, | |
| { | |
| "id": "interactive", | |
| "label": "Interactive", | |
| "count": 11, | |
| "description": "Runtime-update response, state retention, and reuse under prompt updates." | |
| }, | |
| { | |
| "id": "diagnostics", | |
| "label": "Diagnostics", | |
| "count": 4, | |
| "description": "Success- and coverage-adjusted diagnostics that jointly account for update success and source-state coverage." | |
| } | |
| ], | |
| "categories": [ | |
| { | |
| "id": "quality", | |
| "track": "shared", | |
| "label": "Quality & Alignment", | |
| "shortLabel": "Quality", | |
| "description": "Perceptual quality and semantic or temporal correspondence between audio and video.", | |
| "defaultMetric": "va", | |
| "metrics": [ | |
| "va", | |
| "vq", | |
| "pq", | |
| "aq", | |
| "avalign", | |
| "avsync" | |
| ] | |
| }, | |
| { | |
| "id": "streaming", | |
| "track": "shared", | |
| "label": "Continuity & Efficiency", | |
| "shortLabel": "Streaming", | |
| "description": "Technical continuity at native generation boundaries, throughput, and startup latency.", | |
| "defaultMetric": "nbc", | |
| "metrics": [ | |
| "nbc", | |
| "fps", | |
| "ttfc" | |
| ] | |
| }, | |
| { | |
| "id": "adherence", | |
| "track": "progressive", | |
| "label": "Instruction Adherence", | |
| "shortLabel": "Adherence", | |
| "description": "Full-rollout fulfillment and endpoint drift for fixed visual and audio requirements.", | |
| "defaultMetric": "vif", | |
| "metrics": [ | |
| "vif", | |
| "aif", | |
| "vid", | |
| "aid" | |
| ] | |
| }, | |
| { | |
| "id": "stability", | |
| "track": "progressive", | |
| "label": "Long-Horizon Stability", | |
| "shortLabel": "Stability", | |
| "description": "Quality drift, visual consistency, and audio-video drift over the 180-second rollout.", | |
| "defaultMetric": "va_d", | |
| "metrics": [ | |
| "va_d", | |
| "vq_d", | |
| "pq_d", | |
| "aq_d", | |
| "sc", | |
| "bc", | |
| "avalign_d", | |
| "avsync_d" | |
| ] | |
| }, | |
| { | |
| "id": "response", | |
| "track": "interactive", | |
| "label": "Interactive Response", | |
| "shortLabel": "Response", | |
| "description": "Update fulfillment, transition continuity, achievement rate, and successful-response latency.", | |
| "defaultMetric": "vuf", | |
| "metrics": [ | |
| "vuf", | |
| "auf", | |
| "pvc", | |
| "pac", | |
| "pvuar", | |
| "pauar", | |
| "pvrl", | |
| "parl" | |
| ] | |
| }, | |
| { | |
| "id": "state", | |
| "track": "interactive", | |
| "label": "State Retention & Reuse", | |
| "shortLabel": "State", | |
| "description": "Preservation of untargeted states and correct reuse of information established earlier.", | |
| "defaultMetric": "vsr", | |
| "metrics": [ | |
| "vsr", | |
| "asr", | |
| "hdf" | |
| ] | |
| }, | |
| { | |
| "id": "adjusted", | |
| "track": "diagnostics", | |
| "label": "Success & Coverage Adjustments", | |
| "shortLabel": "Adjusted", | |
| "description": "Adjusted diagnostics that penalize unachieved updates and unestablished dependency sources; these are reported separately from the 32 core metrics.", | |
| "defaultMetric": "sa_pvrl", | |
| "metrics": [ | |
| "sa_pvrl", | |
| "sa_parl", | |
| "ca_hdf_a", | |
| "ca_hdf_l" | |
| ] | |
| } | |
| ], | |
| "metrics": { | |
| "va": { | |
| "abbr": "VA", | |
| "name": "Visual Aesthetics", | |
| "direction": "higher", | |
| "range": "[0, 1]", | |
| "unit": "", | |
| "evaluator": "LAION Aesthetic Predictor", | |
| "description": "Perceptual visual appeal, averaged over six 30-second intervals." | |
| }, | |
| "vq": { | |
| "abbr": "VQ", | |
| "name": "Visual Quality", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Visual fidelity, subject integrity, motion naturalness, and artifact control." | |
| }, | |
| "pq": { | |
| "abbr": "PQ", | |
| "name": "Production Quality", | |
| "direction": "higher", | |
| "range": "Original score", | |
| "unit": "", | |
| "evaluator": "Audiobox Aesthetics", | |
| "description": "Perceptual and production fidelity of the generated audio." | |
| }, | |
| "aq": { | |
| "abbr": "AQ", | |
| "name": "Audio Quality", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Audio naturalness and artifact control." | |
| }, | |
| "avalign": { | |
| "abbr": "AVAlign", | |
| "name": "Audio-Video Alignment", | |
| "direction": "higher", | |
| "range": "Cosine similarity", | |
| "unit": "", | |
| "evaluator": "ImageBind", | |
| "description": "Semantic correspondence between generated audio and visual content." | |
| }, | |
| "avsync": { | |
| "abbr": "AVSync", | |
| "name": "Audio-Video Synchronization", | |
| "direction": "lower", | |
| "range": "≥ 0", | |
| "unit": "s", | |
| "evaluator": "Synchformer", | |
| "description": "Absolute predicted temporal offset between audio and video." | |
| }, | |
| "nbc": { | |
| "abbr": "NBC", | |
| "name": "Native Boundary Continuity", | |
| "direction": "higher", | |
| "range": "[0, 100]", | |
| "unit": "", | |
| "evaluator": "Deterministic artifact detector", | |
| "description": "Continuity at model-native generation-unit boundaries, accounting for black frames, repetition, freezing, and flashes." | |
| }, | |
| "fps": { | |
| "abbr": "FPS", | |
| "name": "Frames per Second", | |
| "direction": "higher", | |
| "range": "≥ 0", | |
| "unit": "fps", | |
| "evaluator": "Runtime measurement", | |
| "description": "Generated video frames per unit generation time under the default recommended configuration." | |
| }, | |
| "ttfc": { | |
| "abbr": "TTFC", | |
| "name": "Time to First Chunk", | |
| "direction": "lower", | |
| "range": "≥ 0", | |
| "unit": "s", | |
| "evaluator": "Runtime measurement", | |
| "description": "Elapsed time from receiving a generation request to producing the first output chunk." | |
| }, | |
| "vif": { | |
| "abbr": "VIF", | |
| "name": "Video Instruction Fulfillment", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + case checklist", | |
| "description": "Completeness of visual requirements over the full progressive output." | |
| }, | |
| "aif": { | |
| "abbr": "AIF", | |
| "name": "Audio Instruction Fulfillment", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + case checklist", | |
| "description": "Completeness of audio requirements over the full progressive output." | |
| }, | |
| "vid": { | |
| "abbr": "VID", | |
| "name": "Video Instruction Drift", | |
| "direction": "lower", | |
| "range": "[0, 4]", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + case checklist", | |
| "description": "Absolute change in visual instruction adherence between the first and last intervals." | |
| }, | |
| "aid": { | |
| "abbr": "AID", | |
| "name": "Audio Instruction Drift", | |
| "direction": "lower", | |
| "range": "[0, 4]", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + case checklist", | |
| "description": "Absolute change in audio instruction adherence between the first and last intervals." | |
| }, | |
| "va_d": { | |
| "abbr": "VA-D", | |
| "name": "Visual Aesthetics Drift", | |
| "direction": "lower", | |
| "range": "Absolute VA change", | |
| "unit": "", | |
| "evaluator": "LAION Aesthetic Predictor", | |
| "description": "Absolute change in visual aesthetics between the first and last intervals." | |
| }, | |
| "vq_d": { | |
| "abbr": "VQ-D", | |
| "name": "Visual Quality Drift", | |
| "direction": "lower", | |
| "range": "Absolute VQ change", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Absolute change in visual quality between the first and last intervals." | |
| }, | |
| "pq_d": { | |
| "abbr": "PQ-D", | |
| "name": "Production Quality Drift", | |
| "direction": "lower", | |
| "range": "Absolute PQ change", | |
| "unit": "", | |
| "evaluator": "Audiobox Aesthetics", | |
| "description": "Absolute change in production quality between the first and last intervals." | |
| }, | |
| "aq_d": { | |
| "abbr": "AQ-D", | |
| "name": "Audio Quality Drift", | |
| "direction": "lower", | |
| "range": "Absolute AQ change", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Absolute change in audio quality between the first and last intervals." | |
| }, | |
| "sc": { | |
| "abbr": "SC", | |
| "name": "Subject Consistency", | |
| "direction": "higher", | |
| "range": "Original score", | |
| "unit": "", | |
| "evaluator": "VBench-Long protocol", | |
| "description": "Temporal consistency of subject appearance over the generated sequence." | |
| }, | |
| "bc": { | |
| "abbr": "BC", | |
| "name": "Background Consistency", | |
| "direction": "higher", | |
| "range": "Original score", | |
| "unit": "", | |
| "evaluator": "VBench-Long protocol", | |
| "description": "Temporal consistency of background scenes over the generated sequence." | |
| }, | |
| "avalign_d": { | |
| "abbr": "AVAlign-D", | |
| "name": "Audio-Video Alignment Drift", | |
| "direction": "lower", | |
| "range": "Absolute AVAlign change", | |
| "unit": "", | |
| "evaluator": "ImageBind", | |
| "description": "Absolute change in audio-video semantic correspondence between the first and last intervals." | |
| }, | |
| "avsync_d": { | |
| "abbr": "AVSync-D", | |
| "name": "Audio-Video Synchronization Drift", | |
| "direction": "lower", | |
| "range": "Absolute AVSync change", | |
| "unit": "s", | |
| "evaluator": "Synchformer", | |
| "description": "Absolute change in audio-video synchronization offset between the first and last intervals." | |
| }, | |
| "vuf": { | |
| "abbr": "VUF", | |
| "name": "Video Update Fulfillment", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + update checklist", | |
| "description": "Completeness of each runtime update's visual targets." | |
| }, | |
| "auf": { | |
| "abbr": "AUF", | |
| "name": "Audio Update Fulfillment", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + update checklist", | |
| "description": "Completeness of each runtime update's audio targets." | |
| }, | |
| "pvc": { | |
| "abbr": "PVC", | |
| "name": "Prompt-Based Visual Continuity", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Boundary detector + Gemini 3.1 Pro", | |
| "description": "Visual continuity across prompt-update boundaries while allowing requested changes." | |
| }, | |
| "pac": { | |
| "abbr": "PAC", | |
| "name": "Prompt-Based Audio Continuity", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Audio continuity across prompt-update boundaries while allowing requested changes." | |
| }, | |
| "pvuar": { | |
| "abbr": "PVUAR", | |
| "name": "Prompt-Based Visual Update Achievement Rate", | |
| "direction": "higher", | |
| "range": "[0, 1]", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + update checklist", | |
| "description": "Fraction of applicable visual updates whose core target is clearly realized." | |
| }, | |
| "pauar": { | |
| "abbr": "PAUAR", | |
| "name": "Prompt-Based Audio Update Achievement Rate", | |
| "direction": "higher", | |
| "range": "[0, 1]", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro + update checklist", | |
| "description": "Fraction of applicable audio updates whose core target is clearly realized." | |
| }, | |
| "pvrl": { | |
| "abbr": "PVRL", | |
| "name": "Prompt-Based Visual Response Latency", | |
| "direction": "lower", | |
| "range": "Media-time latency", | |
| "unit": "s", | |
| "evaluator": "Media-time measurement", | |
| "description": "Conditional latency to the first clear realization of an achieved visual target." | |
| }, | |
| "parl": { | |
| "abbr": "PARL", | |
| "name": "Prompt-Based Audio Response Latency", | |
| "direction": "lower", | |
| "range": "Media-time latency", | |
| "unit": "s", | |
| "evaluator": "Media-time measurement", | |
| "description": "Conditional latency to the first clear realization of an achieved audio target." | |
| }, | |
| "vsr": { | |
| "abbr": "VSR", | |
| "name": "Visual State Retention", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Preservation of subject appearance and environment after accounting for requested changes." | |
| }, | |
| "asr": { | |
| "abbr": "ASR", | |
| "name": "Audio State Retention", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Preservation of recurring music, ambience, and persistent sounds not targeted by an update." | |
| }, | |
| "hdf": { | |
| "abbr": "HDF", | |
| "name": "History-Dependency Following", | |
| "direction": "higher", | |
| "range": "1–5", | |
| "unit": "", | |
| "evaluator": "Gemini 3.1 Pro", | |
| "description": "Correct execution of updates using relevant earlier states, evaluated only when the required source state was established." | |
| }, | |
| "sa_pvrl": { | |
| "abbr": "SA-PVRL", | |
| "name": "Success-Adjusted Visual Response Latency", | |
| "direction": "lower", | |
| "range": "[0, 30]", | |
| "unit": "s", | |
| "evaluator": "Achievement-adjusted media-time measurement", | |
| "description": "Visual response latency adjusted for unsuccessful updates." | |
| }, | |
| "sa_parl": { | |
| "abbr": "SA-PARL", | |
| "name": "Success-Adjusted Audio Response Latency", | |
| "direction": "lower", | |
| "range": "[0, 30]", | |
| "unit": "s", | |
| "evaluator": "Achievement-adjusted media-time measurement", | |
| "description": "Audio response latency adjusted for unsuccessful updates." | |
| }, | |
| "ca_hdf_a": { | |
| "abbr": "CA-HDF-A", | |
| "name": "Coverage-Adjusted Adjacent HDF", | |
| "direction": "higher", | |
| "range": "[0, 1]", | |
| "unit": "", | |
| "evaluator": "Source establishment + normalized conditional HDF", | |
| "description": "Adjacent history-dependency quality adjusted for source-state coverage." | |
| }, | |
| "ca_hdf_l": { | |
| "abbr": "CA-HDF-L", | |
| "name": "Coverage-Adjusted Long-Range HDF", | |
| "direction": "higher", | |
| "range": "[0, 1]", | |
| "unit": "", | |
| "evaluator": "Source establishment + normalized conditional HDF", | |
| "description": "Long-range history-dependency quality adjusted for source-state coverage." | |
| } | |
| }, | |
| "models": [ | |
| { | |
| "name": "PixVerse R1", | |
| "paradigm": "Native joint", | |
| "access": "Web", | |
| "reference": "https://pixverse.ai/en/blog/pixverse-launches-r1-real-time-world-model", | |
| "scores": { | |
| "va": 0.529, | |
| "vq": 2.428, | |
| "pq": 6.348, | |
| "aq": 2.81, | |
| "avalign": 0.234, | |
| "avsync": 0.855, | |
| "nbc": null, | |
| "fps": null, | |
| "ttfc": null, | |
| "vif": 2.976, | |
| "aif": 2.511, | |
| "vid": 0.484, | |
| "aid": 0.425, | |
| "va_d": 0.032, | |
| "vq_d": 0.253, | |
| "pq_d": 0.421, | |
| "aq_d": 0.626, | |
| "sc": 0.907, | |
| "bc": 0.93, | |
| "avalign_d": 0.065, | |
| "avsync_d": 0.579, | |
| "vuf": 2.573, | |
| "auf": 2.895, | |
| "pvc": 4.278, | |
| "pac": 2.544, | |
| "pvuar": 0.35, | |
| "pauar": 0.466, | |
| "pvrl": 12.992, | |
| "parl": 10.352, | |
| "vsr": 2.43, | |
| "asr": 4.108, | |
| "hdf": 2.461, | |
| "sa_pvrl": 24.011, | |
| "sa_parl": 20.855, | |
| "ca_hdf_a": 0.232, | |
| "ca_hdf_l": 0.234 | |
| } | |
| }, | |
| { | |
| "name": "HappyOyster", | |
| "paradigm": "Native joint", | |
| "access": "Web", | |
| "reference": "https://www.alizila.com/alibaba-launches-happyoyster-a-world-model-product-for-real-time-immersive-creation-and-interaction/", | |
| "scores": { | |
| "va": 0.529, | |
| "vq": 2.785, | |
| "pq": 6.817, | |
| "aq": 3.157, | |
| "avalign": 0.206, | |
| "avsync": 1.044, | |
| "nbc": null, | |
| "fps": null, | |
| "ttfc": null, | |
| "vif": 3.565, | |
| "aif": 2.692, | |
| "vid": 0.51, | |
| "aid": 0.508, | |
| "va_d": 0.032, | |
| "vq_d": 0.417, | |
| "pq_d": 0.623, | |
| "aq_d": 0.411, | |
| "sc": 0.904, | |
| "bc": 0.934, | |
| "avalign_d": 0.077, | |
| "avsync_d": 0.534, | |
| "vuf": 2.975, | |
| "auf": 2.879, | |
| "pvc": 3.744, | |
| "pac": 3.509, | |
| "pvuar": 0.518, | |
| "pauar": 0.488, | |
| "pvrl": 14.484, | |
| "parl": 13.621, | |
| "vsr": 2.799, | |
| "asr": 3.772, | |
| "hdf": 2.817, | |
| "sa_pvrl": 21.991, | |
| "sa_parl": 21.861, | |
| "ca_hdf_a": 0.354, | |
| "ca_hdf_l": 0.306 | |
| } | |
| }, | |
| { | |
| "name": "OmniForcing", | |
| "paradigm": "Native joint", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2603.11647", | |
| "scores": { | |
| "va": 0.554, | |
| "vq": 2.374, | |
| "pq": 6.351, | |
| "aq": 2.528, | |
| "avalign": 0.119, | |
| "avsync": 1.423, | |
| "nbc": 98.808, | |
| "fps": 12.799, | |
| "ttfc": 3.998, | |
| "vif": 2.423, | |
| "aif": 2.343, | |
| "vid": 0.572, | |
| "aid": 0.277, | |
| "va_d": 0.083, | |
| "vq_d": 0.273, | |
| "pq_d": 0.429, | |
| "aq_d": 0.506, | |
| "sc": 0.962, | |
| "bc": 0.953, | |
| "avalign_d": 0.081, | |
| "avsync_d": 0.479, | |
| "vuf": 1.848, | |
| "auf": 2.242, | |
| "pvc": 4.161, | |
| "pac": 3.369, | |
| "pvuar": 0.053, | |
| "pauar": 0.316, | |
| "pvrl": 13.295, | |
| "parl": 7.925, | |
| "vsr": 2.624, | |
| "asr": 3.388, | |
| "hdf": 2.487, | |
| "sa_pvrl": 29.128, | |
| "sa_parl": 23.046, | |
| "ca_hdf_a": 0.136, | |
| "ca_hdf_l": 0.109 | |
| } | |
| }, | |
| { | |
| "name": "Odyssey-2", | |
| "paradigm": "Cascaded", | |
| "access": "Web", | |
| "reference": "https://odyssey.ml/introducing-odyssey-2", | |
| "scores": { | |
| "va": 0.531, | |
| "vq": 2.82, | |
| "pq": 6.426, | |
| "aq": 2.659, | |
| "avalign": 0.26, | |
| "avsync": 0.946, | |
| "nbc": null, | |
| "fps": null, | |
| "ttfc": null, | |
| "vif": 3.106, | |
| "aif": 2.778, | |
| "vid": 0.5, | |
| "aid": 0.566, | |
| "va_d": 0.04, | |
| "vq_d": 0.463, | |
| "pq_d": 0.362, | |
| "aq_d": 0.456, | |
| "sc": 0.976, | |
| "bc": 0.966, | |
| "avalign_d": 0.064, | |
| "avsync_d": 0.548, | |
| "vuf": 2.408, | |
| "auf": 2.244, | |
| "pvc": 3.919, | |
| "pac": 1.749, | |
| "pvuar": 0.274, | |
| "pauar": 0.308, | |
| "pvrl": 18.758, | |
| "parl": 15.733, | |
| "vsr": 2.836, | |
| "asr": 2.707, | |
| "hdf": 2.554, | |
| "sa_pvrl": 26.838, | |
| "sa_parl": 25.611, | |
| "ca_hdf_a": 0.196, | |
| "ca_hdf_l": 0.168 | |
| } | |
| }, | |
| { | |
| "name": "Self-Forcing", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2506.08009", | |
| "scores": { | |
| "va": 0.585, | |
| "vq": 2.753, | |
| "pq": 6.453, | |
| "aq": 2.623, | |
| "avalign": 0.256, | |
| "avsync": 0.919, | |
| "nbc": 94.359, | |
| "fps": 15.628, | |
| "ttfc": 1.278, | |
| "vif": 3.019, | |
| "aif": 2.81, | |
| "vid": 0.373, | |
| "aid": 0.457, | |
| "va_d": 0.025, | |
| "vq_d": 0.323, | |
| "pq_d": 0.341, | |
| "aq_d": 0.488, | |
| "sc": 0.981, | |
| "bc": 0.969, | |
| "avalign_d": 0.054, | |
| "avsync_d": 0.476, | |
| "vuf": 2.237, | |
| "auf": 2.339, | |
| "pvc": 3.912, | |
| "pac": 1.784, | |
| "pvuar": 0.185, | |
| "pauar": 0.334, | |
| "pvrl": 14.164, | |
| "parl": 12.369, | |
| "vsr": 2.851, | |
| "asr": 2.813, | |
| "hdf": 2.348, | |
| "sa_pvrl": 27.081, | |
| "sa_parl": 24.156, | |
| "ca_hdf_a": 0.206, | |
| "ca_hdf_l": 0.169 | |
| } | |
| }, | |
| { | |
| "name": "LongLive", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2509.22622", | |
| "scores": { | |
| "va": 0.605, | |
| "vq": 2.768, | |
| "pq": 6.459, | |
| "aq": 2.696, | |
| "avalign": 0.272, | |
| "avsync": 0.969, | |
| "nbc": 85.418, | |
| "fps": 10.125, | |
| "ttfc": 1.381, | |
| "vif": 3.164, | |
| "aif": 2.814, | |
| "vid": 0.317, | |
| "aid": 0.581, | |
| "va_d": 0.023, | |
| "vq_d": 0.367, | |
| "pq_d": 0.28, | |
| "aq_d": 0.475, | |
| "sc": 0.986, | |
| "bc": 0.973, | |
| "avalign_d": 0.052, | |
| "avsync_d": 0.476, | |
| "vuf": 2.341, | |
| "auf": 2.138, | |
| "pvc": 3.631, | |
| "pac": 1.809, | |
| "pvuar": 0.241, | |
| "pauar": 0.266, | |
| "pvrl": 11.792, | |
| "parl": 13.078, | |
| "vsr": 2.724, | |
| "asr": 2.712, | |
| "hdf": 2.212, | |
| "sa_pvrl": 25.575, | |
| "sa_parl": 25.49, | |
| "ca_hdf_a": 0.186, | |
| "ca_hdf_l": 0.187 | |
| } | |
| }, | |
| { | |
| "name": "Rolling-Forcing", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2509.25161", | |
| "scores": { | |
| "va": 0.572, | |
| "vq": 2.757, | |
| "pq": 6.393, | |
| "aq": 2.673, | |
| "avalign": 0.265, | |
| "avsync": 1.078, | |
| "nbc": 96.657, | |
| "fps": 10.603, | |
| "ttfc": 2.962, | |
| "vif": 3.009, | |
| "aif": 2.858, | |
| "vid": 0.425, | |
| "aid": 0.474, | |
| "va_d": 0.043, | |
| "vq_d": 0.427, | |
| "pq_d": 0.329, | |
| "aq_d": 0.425, | |
| "sc": 0.98, | |
| "bc": 0.966, | |
| "avalign_d": 0.064, | |
| "avsync_d": 0.506, | |
| "vuf": 2.315, | |
| "auf": 2.221, | |
| "pvc": 4.086, | |
| "pac": 1.8, | |
| "pvuar": 0.217, | |
| "pauar": 0.298, | |
| "pvrl": 13.234, | |
| "parl": 12.82, | |
| "vsr": 2.542, | |
| "asr": 2.597, | |
| "hdf": 2.502, | |
| "sa_pvrl": 26.355, | |
| "sa_parl": 24.941, | |
| "ca_hdf_a": 0.25, | |
| "ca_hdf_l": 0.155 | |
| } | |
| }, | |
| { | |
| "name": "Deep Forcing", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2512.05081", | |
| "scores": { | |
| "va": 0.597, | |
| "vq": 2.632, | |
| "pq": 6.523, | |
| "aq": 2.668, | |
| "avalign": 0.28, | |
| "avsync": 0.878, | |
| "nbc": 92.482, | |
| "fps": 11.065, | |
| "ttfc": 1.397, | |
| "vif": 3.041, | |
| "aif": 2.629, | |
| "vid": 0.359, | |
| "aid": 0.545, | |
| "va_d": 0.023, | |
| "vq_d": 0.334, | |
| "pq_d": 0.289, | |
| "aq_d": 0.497, | |
| "sc": 0.979, | |
| "bc": 0.967, | |
| "avalign_d": 0.062, | |
| "avsync_d": 0.553, | |
| "vuf": 2.295, | |
| "auf": 2.146, | |
| "pvc": 3.879, | |
| "pac": 1.77, | |
| "pvuar": 0.22, | |
| "pauar": 0.288, | |
| "pvrl": 13.433, | |
| "parl": 12.65, | |
| "vsr": 2.449, | |
| "asr": 2.601, | |
| "hdf": 2.206, | |
| "sa_pvrl": 26.42, | |
| "sa_parl": 25.024, | |
| "ca_hdf_a": 0.181, | |
| "ca_hdf_l": 0.15 | |
| } | |
| }, | |
| { | |
| "name": "MemFlow", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2512.14699", | |
| "scores": { | |
| "va": 0.6, | |
| "vq": 2.836, | |
| "pq": 6.407, | |
| "aq": 2.651, | |
| "avalign": 0.269, | |
| "avsync": 1.035, | |
| "nbc": 87.724, | |
| "fps": 10.84, | |
| "ttfc": 1.343, | |
| "vif": 3.142, | |
| "aif": 2.854, | |
| "vid": 0.333, | |
| "aid": 0.492, | |
| "va_d": 0.028, | |
| "vq_d": 0.502, | |
| "pq_d": 0.319, | |
| "aq_d": 0.503, | |
| "sc": 0.986, | |
| "bc": 0.972, | |
| "avalign_d": 0.054, | |
| "avsync_d": 0.474, | |
| "vuf": 2.309, | |
| "auf": 2.286, | |
| "pvc": 3.63, | |
| "pac": 1.88, | |
| "pvuar": 0.217, | |
| "pauar": 0.326, | |
| "pvrl": 13.698, | |
| "parl": 13.154, | |
| "vsr": 2.961, | |
| "asr": 2.879, | |
| "hdf": 2.385, | |
| "sa_pvrl": 26.503, | |
| "sa_parl": 24.433, | |
| "ca_hdf_a": 0.2, | |
| "ca_hdf_l": 0.216 | |
| } | |
| }, | |
| { | |
| "name": "Causal-Forcing", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2602.02214", | |
| "scores": { | |
| "va": 0.556, | |
| "vq": 2.397, | |
| "pq": 6.383, | |
| "aq": 2.714, | |
| "avalign": 0.28, | |
| "avsync": 1.093, | |
| "nbc": 99.558, | |
| "fps": 15.421, | |
| "ttfc": 1.284, | |
| "vif": 2.834, | |
| "aif": 2.729, | |
| "vid": 0.364, | |
| "aid": 0.615, | |
| "va_d": 0.03, | |
| "vq_d": 0.189, | |
| "pq_d": 0.32, | |
| "aq_d": 0.509, | |
| "sc": 0.952, | |
| "bc": 0.946, | |
| "avalign_d": 0.058, | |
| "avsync_d": 0.491, | |
| "vuf": 2.284, | |
| "auf": 2.165, | |
| "pvc": 4.218, | |
| "pac": 1.715, | |
| "pvuar": 0.219, | |
| "pauar": 0.283, | |
| "pvrl": 14.032, | |
| "parl": 12.68, | |
| "vsr": 2.218, | |
| "asr": 2.468, | |
| "hdf": 2.357, | |
| "sa_pvrl": 26.497, | |
| "sa_parl": 25.117, | |
| "ca_hdf_a": 0.184, | |
| "ca_hdf_l": 0.188 | |
| } | |
| }, | |
| { | |
| "name": "Helios", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2603.04379", | |
| "scores": { | |
| "va": 0.47, | |
| "vq": 2.176, | |
| "pq": 6.241, | |
| "aq": 2.521, | |
| "avalign": 0.2, | |
| "avsync": 0.927, | |
| "nbc": 83.657, | |
| "fps": 10.231, | |
| "ttfc": 8.059, | |
| "vif": 2.977, | |
| "aif": 2.712, | |
| "vid": 0.694, | |
| "aid": 0.566, | |
| "va_d": 0.149, | |
| "vq_d": 0.306, | |
| "pq_d": 0.649, | |
| "aq_d": 0.513, | |
| "sc": 0.967, | |
| "bc": 0.967, | |
| "avalign_d": 0.115, | |
| "avsync_d": 0.514, | |
| "vuf": 2.253, | |
| "auf": 2.461, | |
| "pvc": 3.317, | |
| "pac": 1.644, | |
| "pvuar": 0.197, | |
| "pauar": 0.354, | |
| "pvrl": 14.043, | |
| "parl": 12.207, | |
| "vsr": 2.405, | |
| "asr": 1.985, | |
| "hdf": 2.478, | |
| "sa_pvrl": 26.841, | |
| "sa_parl": 23.712, | |
| "ca_hdf_a": 0.202, | |
| "ca_hdf_l": 0.181 | |
| } | |
| }, | |
| { | |
| "name": "SWIFT", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2605.09442", | |
| "scores": { | |
| "va": 0.605, | |
| "vq": 2.735, | |
| "pq": 6.483, | |
| "aq": 2.68, | |
| "avalign": 0.27, | |
| "avsync": 0.97, | |
| "nbc": 87.043, | |
| "fps": 12.359, | |
| "ttfc": 1.496, | |
| "vif": 3.152, | |
| "aif": 2.887, | |
| "vid": 0.305, | |
| "aid": 0.515, | |
| "va_d": 0.022, | |
| "vq_d": 0.411, | |
| "pq_d": 0.272, | |
| "aq_d": 0.463, | |
| "sc": 0.986, | |
| "bc": 0.973, | |
| "avalign_d": 0.049, | |
| "avsync_d": 0.501, | |
| "vuf": 2.357, | |
| "auf": 2.102, | |
| "pvc": 3.615, | |
| "pac": 1.856, | |
| "pvuar": 0.246, | |
| "pauar": 0.276, | |
| "pvrl": 10.34, | |
| "parl": 12.094, | |
| "vsr": 2.609, | |
| "asr": 2.705, | |
| "hdf": 2.322, | |
| "sa_pvrl": 25.15, | |
| "sa_parl": 25.195, | |
| "ca_hdf_a": 0.192, | |
| "ca_hdf_l": 0.232 | |
| } | |
| }, | |
| { | |
| "name": "IAMFlow", | |
| "paradigm": "Cascaded", | |
| "access": "Open", | |
| "reference": "https://arxiv.org/abs/2605.18733", | |
| "scores": { | |
| "va": 0.602, | |
| "vq": 2.84, | |
| "pq": 6.415, | |
| "aq": 2.625, | |
| "avalign": 0.268, | |
| "avsync": 1.016, | |
| "nbc": 86.957, | |
| "fps": 6.049, | |
| "ttfc": 1.772, | |
| "vif": 3.192, | |
| "aif": 2.789, | |
| "vid": 0.302, | |
| "aid": 0.6, | |
| "va_d": 0.023, | |
| "vq_d": 0.425, | |
| "pq_d": 0.331, | |
| "aq_d": 0.563, | |
| "sc": 0.986, | |
| "bc": 0.973, | |
| "avalign_d": 0.057, | |
| "avsync_d": 0.49, | |
| "vuf": 2.319, | |
| "auf": 2.19, | |
| "pvc": 3.655, | |
| "pac": 1.819, | |
| "pvuar": 0.222, | |
| "pauar": 0.297, | |
| "pvrl": 11.683, | |
| "parl": 13.001, | |
| "vsr": 2.836, | |
| "asr": 2.903, | |
| "hdf": 2.424, | |
| "sa_pvrl": 25.976, | |
| "sa_parl": 24.936, | |
| "ca_hdf_a": 0.213, | |
| "ca_hdf_l": 0.183 | |
| } | |
| } | |
| ] | |
| } | |