StreamEval: First Benchmark for Real-Time Interactive Video Generation, 37 Models Evaluated
AdinaYakup · x · 2026-10-22
StreamEval is the first benchmark for interactive streaming video generation, developed with researchers from Tsinghua, NVIDIA, CMU, UC Berkeley, Princeton, NUS, and REACTOR.
- Coverage: 300 test cases (15–300s), 37 models evaluated across five tasks — avatar dialogue, avatar performance, world generation, instruction editing, and reference editing.
- Open & reproducible: All test cases and evaluation code are fully open source, letting anyone inspect and reproduce the results.
- Monthly updates: The team will maintain the leaderboard long-term, adding new models and refreshing results every month.
Related event: StreamEval: First Benchmark for Interactive Streaming Video Generation(2 posts)→
More from Multimodal
- Gradium open-sources Phonon, sub-50M on-device TTS models hitting 1.7% word error rate — mattturck · 2026-10-22
- Synthesia CEO calls Syren "the ChatGPT moment for video" — synthesiaIO · 2026-10-22
- Hermes Agent demo: rigged, dancing game characters from one prompt each, no Blender — alexcovo_eth · 2026-10-22
- AI researcher proposes x0 velocity, a new training target for flow-matching diffusion models — ostrisai · 2026-10-22
- Seedance 2.5 AI-Generated Video Dubbed an 'Absolute Masterpiece' — nikola_mr64990 · 2026-10-22
- LLMSeg: turning off-the-shelf MLLMs into text-prompted segmenters, no training — ducha_aiki · 2026-10-22