NewsFlash Articles Data Fundraising Skill&API

Aethir Mesh Officially Launched: Proprietary Computing Power for Inference, 11 Top Open-Source Models Integrated through Unified API

Read this article in 6 Minutes

Aethir Mesh has launched a total of 11 models from five major open-source providers.

On June 3, 2026, Aethir officially opened Aethir Mesh to all developers and AI agents operators—this is Aethir's self-developed open-source large-scale model API platform, where all inference tasks run directly on Aethir's decentralized GPU infrastructure.

Direct Compute Power Supply, Not Routing Middleware

Understanding the positioning of Aethir Mesh requires first understanding its fundamental difference from existing solutions.

Represented by OpenRouter, mainstream open-source model API platforms are essentially an aggregation routing layer—forwarding user requests to servers of third-party inference providers such as OpenAI, Anthropic, Google, etc., with the platform charging a resale premium. User data inevitably flows through external servers in this process.

Aethir Mesh's architecture is completely different: open models run directly on Aethir's decentralized GPU infrastructure—over 430,000 GPU containers spanning 94 countries and 200+ nodes. Aethir is the layer actually performing the computation, rather than a billing wrapper for others' APIs. This means no resale premium, no third-party data routing, and enterprise-level service standards with GPU utilization remaining above 95%.

11 Models, One Endpoint

Aethir Mesh has currently launched a total of 11 models from five major open-source providers, with a core lineup that includes:

• DeepSeek V4: A 1.6 trillion-parameter MoE architecture where only a small part is activated per inference step, achieving significantly lower than market-priced inference costs approaching the performance of cutting-edge proprietary models, suitable for most agent general tasks;

• Kimi K2.6: Produced by Moonshot AI, a 1 trillion-parameter MoE with 320 billion active parameters per step, natively supporting large-scale sub-agent collaborative orchestration, designed for multi-agent pipeline scenarios;

• GLM-5.1: Produced by Z.AI, under the MIT open-source license, MoE architecture with 400 billion active parameters per step, a super long-context window designed for long-range tasks, supporting single continuous runs lasting up to 8 hours and thousands of tool invocations;

• MiniMax-M2.5: Released in February 2026, aimed at programming, agent tool use, web search, and complex office task scenarios, achieving an inference speed of 206.7 tokens per second;

• Qwen3.6-27B: Developed by Alibaba, with 27 billion parameters, Apache 2.0 license, native 262K Token context (scalable to millions), and a comprehensive performance improvement over the previous flagship Qwen3.5-397B.

All models are accessed through a unified API endpoint, allowing developers to manage a single key, a single bill, and a single dashboard without the need to maintain multiple sets of authentication processes across platforms.

Integration with Aethir Claw

For users of the Aethir Claw intelligent agent platform, Aethir Mesh is the final piece of their full-stack closed loop. After users write the Mesh API Key into the intelligent agent configuration on the Aethir Claw dashboard, all the intelligent agent's model inference requests will automatically be routed through Aethir Mesh and the results returned within the platform—VPS instances, skill execution layer, and model inference layer all share the same infrastructure, unified billing, and monitoring, with the inference data never leaving the Aethir ecosystem.

For independent developers and businesses not using Aethir Claw, Aethir Mesh can also serve as a standalone inference endpoint, suitable for Agentic AI application development, DePIN projects, and enterprise-grade AI workflows.

Expanding Model Catalog

Aethir Mesh's model catalog is designed to dynamically expand with the open-source ecosystem. As high-performance open-source models are continuously released and reach production stability, Mesh will consistently include new models, allowing users to access the latest capabilities without switching API providers or renegotiating access terms.

Technical Documentation: docs.aethir.com/aethir-mesh

Get Started Now: https://agent.aethir.com/twitterCNmesh