StreamingMemeStreamingMeme
LeaderboardsEventsSubmit News
SUBSCRIBE

Daily Brief

The streaming industry in your inbox every morning.

Daily Brief

The streaming industry in your inbox every morning.

StreamingMeme

The streaming technology industry news aggregator.

About UsNewsletterSubmit News
© 2026 StreamingMeme. All rights reserved.
← AI for Video
AI & VideoTechnical DevelopmentMay 18, 2026

DeltaToken cuts video tokens from 180K to under 1,000

Qiang Zhang

Qiang Zhang announced 'DeltaToken', a new video tokenizer designed to reduce the number of VAE tokens for video models by up to 192x while maintaining the same number of channels. This advancement is stated to lower training costs, increase inference savings for real-time video generation, and extend video context length from seconds to minutes for AI models.

Key Takeaways

  • DeltaToken is a new video tokenizer for world models and video models that uses the same number of channels while cutting VAE tokens by up to 192x.
  • One example in the post shows token count falling from 180K to under 1,000.
  • The project claims 10–100x lower training cost, with a video foundation model trained from scratch for under $4,000 in compute.
  • The post says the compression could extend context length from 10–15 seconds to 5–10 minutes for native cross-shot consistency.
  • Qiang Zhang says the encoder focuses on what changes in video, which he says improves physical grounding for embodied world models.

Why It Matters

If the claims hold up, DeltaToken reduces the token burden that sits between raw video and model training, inference, and longer-context generation. That matters most for systems trying to run video generation in LLMs, VLMs, and VLAs, since the post argues the compression makes native integration possible without architectural compromise. The immediate technical signal is cost: sub-$4,000 scratch training and real-time on-device generation are both called out. Watch for the released demo details and whether the 180K-to-under-1,000 token reduction holds across different video workloads.


Read full article at linkedin.com

Related Articles

Broadcast: AMD pushes AI to the edge for live broadcast latency and trust
Spotify Engineering: Spotify: 99% of Engineers Use AI Coding Tools Weekly, Productivity Up 76%
Startuphub: Wasmer builds Node.js edge runtime in two weeks using OpenAI Codex

Newest

in about 1 hour
The Broadcast Bridge: Decoding H.264: Navigating AVC Profiles, Levels, and Signaling for Streaming
about 10 hours ago
Valkey: Momento CTO Details Valkey's Role in High-Scale Streaming and AI Caching
about 10 hours ago
Cloudinary: Cloudinary Publishes Guide for Migrating Media Assets to Its Platform
about 10 hours ago
Azure Player: AzurePlayer Updates Focus on Video Playback Optimization for Streaming Professionals
about 10 hours ago
Upwork: Upwork Spotlights GLSL Specialists for Video Processing and Edge AI
about 10 hours ago
F6s: Global-M Platform Monetizes Video Content for Over 25 Operators
about 10 hours ago
Broadcastbeat: Mediaproxy LogServer Adds Nielsen CBET for Radio Audience Measurement
about 10 hours ago
Senza Fili: Wi-Fi Alliance Certifies 6 GHz Wi-Fi 6E, Boosting Streaming Throughput and Latency
about 10 hours ago
Span: XFRA to Deploy 1 Gigawatt of AI Compute in Homes by 2027
about 10 hours ago
Intelmarketresearch: Edge Computing Market to Exceed $31B by 2034, Driven by 5G and Immersive Media
about 10 hours ago
Fierce Network:
about 10 hours ago
SQLServerCentral: Edge AI is for Constraints, Not for Aesthetics, New Report Warns
about 10 hours ago
YouTube: YouTube Shorts Launches AI 'Dream Screen' for Background Generation
about 10 hours ago
Network World: Edge Computing, Private 5G Cut Live Event Latency to Under 5ms
about 10 hours ago
YouTube: Canada Reverses 15% Streaming Revenue Requirement Amid U.S. Trade Pressure
about 10 hours ago
YouTube: Canada Backs Off CRTC's Triple Fee Hike on Streamers Amid US Trade Concerns
about 10 hours ago
In: Amazon Plans Indiana Data Center Expansion, Faces Environmental Review
about 10 hours ago
BeBee: Hearst's WTAE-TV Seeks Digital Sales Manager to Drive AI-Enhanced Revenue
about 11 hours ago
LinkedInEditors: NETINT: VPU performance hinges on system architecture, not just silicon efficiency
about 11 hours ago
Microsoft: Microsoft Foundry Unveils MAI-Voice-2 AI for Multilingual Speech Generation

Upcoming Events

Jun
8–11
NEM Dubrovnikhttps://neweumarket.com/dubrovnik/
Jun
11–12
Arctic 15https://arctic15.com/
Jun
13–19
InfoCommhttps://www.infocommshow.org/
Jun
16–19
Stream TV Show (formerly the Pay TV Show)https://www.streamtvshow.com/
Jun
17–19
Content Tokyo 2024https://www.content-tokyo.jp/ja-jp.html
View all events →

Top Sources

  1. 1.MSN155
  2. 2.Calendly106
  3. 3.Sports Video Group65
  4. 4.Advanced Television63
  5. 5.TV Technology42
  6. 6.Cord Cutters News41
  7. 7.Broadband TV News36
  8. 8.AOL34
Full leaderboards →

Newest

in about 1 hour
The Broadcast Bridge: Decoding H.264: Navigating AVC Profiles, Levels, and Signaling for Streaming
about 10 hours ago
Valkey: Momento CTO Details Valkey's Role in High-Scale Streaming and AI Caching
about 10 hours ago
Cloudinary: Cloudinary Publishes Guide for Migrating Media Assets to Its Platform
about 10 hours ago
Azure Player: AzurePlayer Updates Focus on Video Playback Optimization for Streaming Professionals
about 10 hours ago
Upwork: Upwork Spotlights GLSL Specialists for Video Processing and Edge AI
about 10 hours ago
F6s: Global-M Platform Monetizes Video Content for Over 25 Operators
about 10 hours ago
Broadcastbeat: Mediaproxy LogServer Adds Nielsen CBET for Radio Audience Measurement
about 10 hours ago
Senza Fili: Wi-Fi Alliance Certifies 6 GHz Wi-Fi 6E, Boosting Streaming Throughput and Latency
about 10 hours ago
Span: XFRA to Deploy 1 Gigawatt of AI Compute in Homes by 2027
about 10 hours ago
Intelmarketresearch: Edge Computing Market to Exceed $31B by 2034, Driven by 5G and Immersive Media
about 10 hours ago
Fierce Network:
about 10 hours ago
SQLServerCentral: Edge AI is for Constraints, Not for Aesthetics, New Report Warns
about 10 hours ago
YouTube: YouTube Shorts Launches AI 'Dream Screen' for Background Generation
about 10 hours ago
Network World: Edge Computing, Private 5G Cut Live Event Latency to Under 5ms
about 10 hours ago
YouTube: Canada Reverses 15% Streaming Revenue Requirement Amid U.S. Trade Pressure
about 10 hours ago
YouTube: Canada Backs Off CRTC's Triple Fee Hike on Streamers Amid US Trade Concerns
about 10 hours ago
In: Amazon Plans Indiana Data Center Expansion, Faces Environmental Review
about 10 hours ago
BeBee: Hearst's WTAE-TV Seeks Digital Sales Manager to Drive AI-Enhanced Revenue
about 11 hours ago
LinkedInEditors: NETINT: VPU performance hinges on system architecture, not just silicon efficiency
about 11 hours ago
Microsoft: Microsoft Foundry Unveils MAI-Voice-2 AI for Multilingual Speech Generation

Upcoming Events

Jun
8–11
NEM Dubrovnikhttps://neweumarket.com/dubrovnik/
Jun
11–12
Arctic 15https://arctic15.com/
Jun
13–19
InfoCommhttps://www.infocommshow.org/
Jun
16–19
Stream TV Show (formerly the Pay TV Show)https://www.streamtvshow.com/
Jun
17–19
Content Tokyo 2024https://www.content-tokyo.jp/ja-jp.html
View all events →

Top Sources

  1. 1.MSN155
  2. 2.Calendly106
  3. 3.Sports Video Group65
  4. 4.Advanced Television63
  5. 5.TV Technology42
  6. 6.Cord Cutters News41
  7. 7.Broadband TV News36
  8. 8.AOL34
Full leaderboards →