Domestic Agent model “Kunlun Wanwei” breaks into the global top tier!

Limited-time free—performance has reached the level of closed-source/top proprietary leaders, and the price has been cut to the industry mid-range; after that, there’s also an open-source backup.

SkyClaw-v1.0, the main force for handling complex tasks.
It has been deeply optimized for OpenClaw-related tasks and is suitable for enterprise-grade multi-step scenarios with heavy tool dependencies.
The lightweight version SkyClaw-v1.0-lite doesn’t compromise the core capabilities of the Agent, but it’s faster and cheaper.
It’s aimed at high-frequency, cost-sensitive scenarios, such as bulk API calls and lightweight automation workflows.
SkyClaw-v1.0 input is 0.5 yuan per million tokens, output is 4 yuan per million tokens; the lightweight version’s input is only 0.3 yuan, and output is 2 yuan.
As early as February this year, Kunlun Wanwei launched the SkyClaw product, rebuilt based on the open-source architecture OpenClaw, delivering an always-on cloud AI assistant.
In the following March, it also rolled out its first batch of 6 official Skills on GitHub and Clawhub: PPT, documents, spreadsheets, design, search, and music—essentially covering the core links of office scenarios.
Now that SkyClaw-v1.0 is out, looking back at this line, it’s crystal clear:
First build the platform, then expand the Skill ecosystem, and finally solve the performance and cost of the underlying models together.
Now that SkyClaw-v1.0 is out, looking back at this line, it’s crystal clear:
First build the platform, then expand the Skill ecosystem, and finally solve the performance and cost of the underlying models together.