Unverified50% confidenceFactExact time
FastH3 V2 仅需 8 步采样即可完成视频生成
1
Sources
50%
Confidence
Long-term
Relevance
9/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMiniMax H3模型在vLLM-Omni推理框架和FastH3加速方案支持下，用8.7秒完成了10.1秒含画面与同步音频的MP4视频生成71% similarUnverifiedFastH3 V2 是 FastVideo 推出的视频扩散模型的最新迭代版本69% similarUnverified少步数视频生成背后的核心技术通常是一致性蒸馏或流匹配加速63% similarUnverified生成一段1080P、10秒左右的视频耗时约5分钟，单条视频花费不到3毛钱63% similarUnverifiedFastVideo项目专注于视频生成的推理加速，目标是让视频生成更快、更省63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/926163API
curl https://kongchang.com/api/v1/knowledge/claims/926163MCP
get_claim(id=926163)