Introducing 0GM-1.0-35B-A3B. Our first proprietary AI model. Mixture of Experts (MoE), 35B parameters, 3B active per token. Trained on our own decentralized GPU network. Open source under Apache 2.0.

0G to Make Alibaba's Qwen wModels Accessible to AI Agents via Blockchain Integration SINGAPORE, April 21, 2026 — AI agents can now directly access top-tier large language models on-chain for the first time. A new collaboration between 0G Foundation and Alibaba Cloud, the digital and intelligence backbone of Alibaba Group, enables direct, blockchain-based access to Alibaba's Qwen — one of the world’s most widely adopted large language model families. This collaboration marks a shift from API-based AI access to programmable, tokenized AI infrastructure. AI Agents Lack Seamless, Scalable Access to Top-Tier Models AI agents — software programs capable of autonomously querying models, making decisions, and executing workflows — are gaining traction across industries. However, their effectiveness is still constrained by how easily they can access leading AI models. Today, most large language models remain gated behind centralized APIs, requiring account setup, fiat-based billing, and manual configuration. These systems were not designed for autonomous, high-frequency agent interactions, where seamless, real-time access is critical. This creates a structural limitation: while models like Qwen are powerful, integrating them into agent-driven environments at scale remains complex and inefficient. Direct On-Chain Access to Qwen Through this collaboration, 0G enables developers to access Qwen’s latest models in a way that is inherently compatible with autonomous systems. By procuring Qwen tokens via API and embedding access into its infrastructure, 0G allows AI agents to invoke Qwen models directly using a token-based mechanism. This removes the need for traditional account management and enables programmatic, on-demand access to AI compute. In practical terms, AI agents operating on 0G can query Qwen to generate responses, process information, or execute tasks — creating a more fluid interaction between decentralized systems and state-of-the-art AI models. At a high level, this collaboration reflects a clear division of roles in the emerging AI agent stack: Qwen powers intelligence, while 0G guarantees trust. In practical terms, inference runs on Qwen, and verification runs on 0G — forming a more complete foundation of compute and trust for autonomous AI systems. As Michael Heinrich, CEO and Co-founder of 0G Labs, put it: “As the world’s leading Blockchain for AI Agents, we are delighted to become a preferred partner for Alibaba’s Qwen models. 0G’s integration with Alibaba Cloud’s Qwen marks a milestone where 'top-tier intelligence meets trusted infrastructure.' With Qwen 3.6-Plus optimized for agentic AI, Alibaba provides powerful reasoning capabilities while 0G provides the verifiable trust layer, together building the most complete trust loop for AI Agents. We look forward to consistently deepening our collaboration with Alibaba Cloud and the Qwen team to accelerate the mass adoption of AI Agents globally.” Why Qwen Matters The collaboration underscores Qwen’s growing role as a defining force in the global AI ecosystem. Developed by Alibaba, Qwen has become one of the most widely downloaded model families, supporting a broad range of multimodal and enterprise use cases. Its latest release, Qwen3.6, has demonstrated strong performance across multiple benchmarks, reinforcing its position among the world’s leading AI systems. As competition intensifies among leading AI models, accessibility and distribution are becoming as critical as model performance. Enabling AI agents to interact with Qwen through decentralized infrastructure extends its reach beyond traditional enterprise environments into emerging autonomous systems. Enabling Real AI Agent Workflows The ability of AI agents to directly access Qwen unlocks new possibilities for real-world applications. Developers can now build systems where agents: ● Query Qwen for natural language understanding and generation ● Automate decision-making processes ● Execute multi-step workflows powered by AI reasoning By removing access friction, the collaboration shifts the focus from infrastructure constraints to application development, enabling more rapid experimentation with agent-based systems. Together with 0G’s verifiable compute and data layer, this significantly enhances trust and auditability in agent-driven environments — a key requirement for large-scale autonomous systems.