AIOZ Network Report: September 2026
September expanded the AIOZ Network ecosystem across both DePIN-powered storage and AI, with new AIOZ Storage capabilities for AI-native workflows alongside continued growth in AIOZ AI models, datasets, and challenges.
From vector-based retrieval and persistent agent memory to new models, datasets, and challenges, this report highlights the key updates from September.
AIOZ Storage Updates
Vector Database
AIOZ Storage introduced Vector Database, extending its storage infrastructure to support vector-based retrieval for AI applications.
Vector databases enable applications to organize and retrieve information based on semantic similarity, creating a practical foundation for workflows such as semantic search, retrieval-augmented generation, recommendation systems, and AI agents that need fast access to relevant context.
Discover the update:
Agent Memory
AIOZ Storage Agent Memory gives AI agents a persistent memory layer through a simple Record and Recall workflow.
Record stores a statement verbatim, while Recall retrieves relevant memories based on meaning and returns a similarity score. Memories can be returned exactly as recorded, recorded immediately without polling, and retrieved across languages.
Discover the update:
AIOZ AI Ecosystem Expansion
New Models on AIOZ AI
MiMo-7B-Base is a 7B-parameter base language model designed as a flexible foundation for language understanding, generation, and downstream model development.
Ministral 3 3B Base brings a compact model profile for developers exploring efficient language-model workflows. Its smaller scale makes it relevant for applications where deployment efficiency and general language capabilities need to work together.
Qwen3-0.6B is a compact language model designed for conversational AI, instruction following, reasoning, and multilingual applications.
Qwen2.5-72B-Instruct expands the model library with a large instruction-tuned language model built for complex language understanding, generation, and instruction-following workflows.
InsectSAM is an insect segmentation model designed to identify and segment insects from complex backgrounds, supporting computer vision workflows for insect detection and biodiversity monitoring.
New AI Datasets on AIOZ AI
DocLayNet is a dataset designed for document layout analysis, supporting workflows in document understanding, layout detection, and information extraction.
SciTLDR focuses on scientific text summarization, supporting the development and evaluation of models designed to understand technical content and produce compact representations of scientific research.
New AIOZ AI Challenge
The Handwritten Digit Recognition Challenge is live now, introducing a classic computer vision classification task.
Participants build models to identify handwritten digits from image data, exploring fundamental workflows across image preprocessing, classification, model evaluation, and optimization.
The challenge provides an accessible environment for testing computer vision approaches while building practical experience with supervised machine learning.
Join the challenge:
What’s Next
September expanded AIOZ Network across both storage infrastructure and AI resources. With Vector Database and Agent Memory supporting AI-native data workflows, alongside new models, datasets, and challenges on AIOZ AI, the ecosystem continues to support a broader range of practical applications built on DePIN.
Explore more and start building on AIOZ.
About the AIOZ Network
AIOZ Network is a DePIN for AIOZ AI, AIOZ Storage, AIOZ Pin and AIOZ Stream.
Powered by a global community of AIOZ DePINs, AIOZ rewards you for sharing your computational resources for storing, transcoding, and streaming digital media content and powering decentralized AI computation.
Find Us
AIOZ All Links | Website | X | Telegram