Close Menu
CrypThing
  • Directory
  • News
    • AI
    • Press Release
    • Altcoins
    • Memecoins
  • Analysis
  • Price Watch
  • Price Prediction
Facebook X (Twitter) Instagram Threads
CrypThingCrypThing
  • Directory
  • News
    • AI
    • Press Release
    • Altcoins
    • Memecoins
  • Analysis
  • Price Watch
  • Price Prediction
CrypThing
Home»News»Tether Data Pushes AI Off the Cloud With New 460M Parameter Vision Model
News

Tether Data Pushes AI Off the Cloud With New 460M Parameter Vision Model

adminBy adminJuly 29, 20263 Mins Read
Share Facebook Twitter Pinterest LinkedIn Tumblr Email Copy Link Bluesky Reddit Telegram WhatsApp Threads
Tether Data Pushes AI Off the Cloud With New 460M Parameter Vision Model
Share
Facebook Twitter Email Copy Link Bluesky Reddit Telegram WhatsApp

Key Takeaways

Tether’s QVAC unit open-sourced a 460-million-parameter vision model on July 29, 2026.The model beat Liquid AI and Hugging Face rivals across 16 of 17 benchmark tests.Its Flash version ran up to 36 times faster than SmolVLM2-500M on an iPhone 15.

Released Wednesday by the company’s AI research arm, QVAC, VisionPsy-Nano can examine images, documents, and charts, then answer questions without sending the source material to a remote system. The phone handles both the input and response, allowing the software to operate offline and keep data on the device.

QVAC says Tether AI Research’s model recorded the highest overall score among vision-language systems with fewer than 500 million parameters. Across 17 benchmarks, it beat competing models on 16.

How It Stacks Up

The model posted a normalized score of 62.3, compared with 59.6 for Liquid AI’s LFM2.5-VL-450M and 52.5 for Hugging Face’s SmolVLM2-500M. QVAC’s earlier nanoVLM-460M-8k base model scored 54.9.

The Tether Data release comes in two versions. The standard build prioritizes accuracy, while Flash gives up a small amount of quality for faster responses. QVAC says Flash retains about 99% of the full model’s performance while producing its first output up to 23 times faster than SmolVLM2-500M on Android phones and up to 36 times faster on an iPhone 15.

That response time matters on mobile hardware. A compact model can score well in testing but still feel impractical if users must wait while the device processes an image. Flash is designed to reduce that initial delay.

The artificial intelligence (AI) system also supports document analysis and optical character recognition, extracting text and structure from financial reports, flowcharts and infographics.

QVAC says the model beat similarly sized competitors by an average of 7.4% on visual reasoning tests. It also outperformed models more than twice its size on MM-IFEval and POPE, including releases from Qwen and InternVL.

Why the Small Footprint Matters

Moving image processing from a data center to a phone introduces strict limits around memory, heat, battery use, and computing power. Compact models often handle plain text but lose accuracy when reading dense charts, tables, or scanned documents.

QVAC’s benchmark results suggest the model preserved much of that capability while remaining small enough for consumer devices. Local processing also means documents do not need to be uploaded, and the software can continue working without a network connection.

Built for Several Deployment Options

Developers can access the model through Hugging Face Transformers, a quantized GGUF version for llama.cpp or a vLLM backend for higher-volume server use.

QVAC also published its benchmark configurations through VLMEvalKit, allowing outside researchers to repeat the tests. Both versions use the Apache 2.0 license, permitting commercial use, modification, and redistribution.

Part of Tether’s Wider AI Strategy

Tether CEO Paolo Ardoino said the release supports the company’s push toward local, efficient AI systems.

“Achieving best-in-class quality and performance on general vision tasks at just 460 million parameters proves that local-first, highly efficient AI is a viable pathway,” Ardoino said.

The model follows several QVAC releases dating to October 2025, including synthetic training datasets, a cross-platform software development kit, and medical models designed for phones and wearable devices.

For developers, the release offers offline image analysis without usage-based API costs. Businesses can keep sensitive documents on the device, while consumers can use AI features without a signal. Researchers can independently test the company’s performance claims using the published configurations.

Tether has financed its AI expansion with profits from its USDT stablecoin business. It has also invested in AI infrastructure, biotechnology, and robotics, including a Series C investment in NEURA Robotics valued at up to $1.4 billion.

460M Cloud Data model Parameter pushes Tether vision
Share. Facebook Twitter Pinterest LinkedIn Tumblr Telegram Email Copy Link Bluesky WhatsApp Threads
Previous ArticleZuckerberg says Meta’s enterprise AI opportunity extends beyond agents
Next Article AMD Rethinks Memory Strategy with EPYC 9004, 9005 CPUs
admin

Related Posts

Morgan Stanley Launches Ether and Solana ETPs With Staking and Market-low Fees

July 28, 2026

KB Kookmin Taps JPMorgan’s Kinexys as South Korea’s Largest Bank Pushes Instant Dollar Payments

July 27, 2026

Physicists Recreated the Way a Spinning Black Hole Leaks Energy, on a Tabletop

July 26, 2026
Trending News

Bitmine Immersion Technologies (BMNR) Announces ETH Holdings Reach 5.79 Million Tokens, and Total Crypto and Total Cash Holdings of $11.8 Billion

July 27, 2026

What Makes a Casino Truly Crypto-First? A Look Beyond Bitcoin Payments

July 23, 2026

HTX Growth Academy Analyzes 2025 Crypto Market Shifts and 2026 Projections

December 29, 2025

Trump could soon sign executive order to penalize banks for discriminating against crypto firms

August 5, 2025
About Us

At crypthing, we’re passionate about making the crypto world easier to (under)stand- and we believe everyone should feel welcome while doing it. Whether you're an experienced trader, a blockchain developer, or just getting started, we're here to share clear, reliable, and up-to-date information to help you grow.

Don't Miss

Reporters found that Zerebro founder was alive and inhaling his mother and father’ home, confirming that the suicide was staged

May 9, 2025

Openai launches initiatives to spread democratic AI through global partnerships

May 9, 2025

Stripe announces AI Foundation model for payments and introduces deeper Stablecoin integration

May 9, 2025
Top Posts

Bitmine Immersion Technologies (BMNR) Announces ETH Holdings Reach 5.79 Million Tokens, and Total Crypto and Total Cash Holdings of $11.8 Billion

July 27, 2026

What Makes a Casino Truly Crypto-First? A Look Beyond Bitcoin Payments

July 23, 2026

HTX Growth Academy Analyzes 2025 Crypto Market Shifts and 2026 Projections

December 29, 2025
  • About Us
  • Privacy Policy
  • Terms and Conditions
  • Disclaimer
© 2026 crypthing. All Rights Reserved.

Type above and press Enter to search. Press Esc to cancel.