viraldi.
Video details
How it works
Back to discovery
14:26
Technology & Computing / VIDEO

Xiaomi & DeepSeek Just Solved the Biggest Bottleneck in Your Local AI

Kai· 58.4K subscribers

23d ago · English audio

Xiaomi MiMo V3's HySparse2 and DeepSeek V4.1 Flash both shrank the KV cache, and both papers say memory was never the main problem. On September 10th DeepSeek shipped V4.1 Flash, which holds a million tokens of context in 0.93 GB of KV cache: 890 bytes per token. On September 22nd Xiaomi published HySparse2, the architecture Fuli Luo says is the core of MiMo V3, which cuts KV cache 4.5 times against the Hybrid SWA…

Watch on YouTube
6.3×Above the channel median
118.1KViews
3.6Avg. views / hour · 1.1h window
18.8KMedian · 50 videos

Outlier score

118.1K views ÷ a 18.8K baseline = 6.3×. Compared with 50 earlier videos from the same channel, matching the format recorded at scoring time. Older-upload baseline: reference videos had at least as long to gain views. This comparison is not age-matched.

YouTube snapshot: Sun, 25 Oct 2026 12:33:31 GMT. Counters may have changed.

Read the methodology

Generative AI & Agents

Channel-balanced percentile: 98

Comparison details

45 videos from 30 other channels, with each channel weighted equally. Same format and language, with 10,000–99,999 subscribers. Observed at 573 hours after publication, within ±48 hours.

This describes the collected sample, not all of YouTube. Updated Sun, 25 Oct 2026 13:31:01 GMT.

Niches & tags

Transcripts and detailed analytics are available for recent outlier Shorts up to three minutes. Research tools

Breakouts from other channels