Apple open-sources FastFrames real-time video models with 36 ms single-GPU latency
11/18/2026 — 11/18, 18:21·1 sources (all)·1 reports (all)
Key takeaways
- Apple open-sourced FastFrames, a family of video generation models built for real-time interaction: the 1.3B-parameter FastFrames-S and 14B-…
Background & analysis
On November 18, 2026, Zhidongxi reported that Apple had open-sourced the FastFrames video generation model family, a series built for real-time interactive use. The release includes two models: FastFrames-S with 1.3B parameters and FastFrames-L with 14B parameters, both aimed at video generation researchers and developers building interactive applications, with a focus on low-latency 480P video output.
According to testing data released by Apple, FastFrames-S generates video with a latency of 36 ms on a single NVIDIA B200 GPU, roughly 7.6x faster than its foundation model, Causal Forcing++, while keeping generation quality at a comparable level. FastFrames-L matches the output quality of Krea-Realtime, another real-time video model also built on Wan2.1-14B, while running about 10.6x faster on a single card; when scaled to 8 B200 GPUs, its latency drops further to 45 ms.
To check whether the same acceleration approach could be extended to other models, Apple also optimized two external models in its tests. LingBot 1.3B, from Ant Group's Lingbo team, posted a response speed roughly 16x that of the original after optimization, while RealWonder, from a team at Stanford University and collaborators, reached about 22x its original response speed. Public reporting has so far covered only these test results, and no further developments have been disclosed.
AI-generated from 1 reports · updated 4 days ago
- Heat index
- 25
- All sources
- 1
- All reports
- 1
- First seen
- 4 days ago
Reports on this story headlines open the original
Apple has open-sourced FastFrames, a family of real-time interactive video generation models with 1.3B and 14B parameter versions targeting low-latency 480P output. Apple says FastFrames-S generates at 36ms latency on a single NVIDIA B200, roughly 7.6x faster than its base model Causal Forcing++.
智东西AI score 59
Other stories people are talking about
- Heat index 1143RisingAnthropic Reaches $1.6 Trillion Market Value in First Trading Week After IPO21 sources
- Heat index 1011Anthropic Mathematician Says Claude Formalized the Generalized Riemann Hypothesis21 sources
- Heat index 622Anthropic IPO total raise reaches $93.5 billion after greenshoe exercise20 sources
- Heat index 596RisingThinking Machines Lab Open-Sources Inkling 2 Mini with 32GB Model Card Recommendation10 sources
- Heat index 490RisingSEC probes Nvidia accounting practices as shares post seven-week slide9 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

