Aethir ships Claw agent platform

Aethir said on May 13 that it had added a native large language model API layer to Aethir Claw, turning the product into a more complete stack for deploying AI agents. The company said the new Model-as-a-Service, or MaaS, layer bundles model access with the managed OpenClaw hosting product it introduced in alpha on March 25. The pitch is straightforward: users no longer need to separately buy a virtual private server, manage external API keys and reconcile multiple bills across different vendors. Aethir said Claw now offers access to models from Anthropic, OpenAI and Google inside the same subscription. ### What exactly changed in Claw this week? Aethir’s May 13 post said the biggest change was the launch of a native MaaS layer inside Claw. The company described that as bundled LLM API credits that remove the need for separate provider accounts, separate API keys and a second bill for model usage. (ecosystem.aethir.com) The company’s earlier March 25 launch described Claw as an agent-hosting platform built on Aethir’s decentralized GPU cloud infrastructure. At that stage, the emphasis was on managed deployment, isolated VPS instances and optional “zero-provider-access” controls rather than integrated model access. ### What does Aethir say users get in one package? (ecosystem.aethir.com) Aethir’s documentation says Claw is a browser-based managed platform where users can deploy a fully configured OpenClaw agent instance with one click. The docs say the service includes the agent, the compute, the models and the dashboard, with no terminal or local-machine setup required. Aethir’s product materials say each instance runs in a fully isolated environment with Ubuntu 24.04 LTS, Google Chrome pre-installed for browser automation and OpenClaw ready to launch. (blog.aethir.com) The company also says users can choose deployment regions in the United States, Japan or Singapore. ### How is the company framing the problem it wants to solve? (docs.aethir.com) Aethir said on May 13 that deploying an AI agent typically means stitching together a VPS or cloud server, an LLM API provider, API key management and separate billing. The company said that fragmentation is a hurdle for users who want to deploy agents without handling infrastructure work themselves. (blog.aethir.com) The March 25 launch post made the same case in more operational terms. Aethir said OpenClaw setup can involve Docker, SSH, VPS configuration, terminal commands and ongoing server maintenance, and it argued that many users want a shorter path from signup to a working agent. ### What infrastructure is Claw built on? (ecosystem.aethir.com) Aethir says Claw runs on its own decentralized GPU cloud infrastructure rather than on resold third-party virtual servers. The company’s main site says Aethir has more than 435,000 enterprise-grade GPUs on demand, more than $400 million worth of compute capacity and support across 93 countries. (blog.aethir.com) Aethir’s docs say that vertical integration is central to the product. The company says Claw combines agent hosting and model hosting for inference on a single platform using GPU infrastructure supplied by decentralized cloud hosts. ### What comes next for Claw? Aethir’s May 13 post said the current MaaS layer consolidates frontier and open-source model access under one subscription today. (blog.aethir.com) The company said its roadmap includes hosting leading models directly on Aethir’s decentralized GPU network in the future. (docs.aethir.com) The company has also been broadening the product around packaged agent use cases. An April 17 post said Claw was moving toward easier browser-based deployment, and later materials described prebuilt crypto-focused agents running on the same infrastructure. Aethir’s latest public milestone is the May 13 MaaS release, and the next named step in its own roadmap is direct hosting of leading models on Aethir GPUs, according to the company’s announcement. (ecosystem.aethir.com) (blog.aethir.com)