Tencent (腾讯) explains “Lobster Family” OpenClaw debate: installation is free, model calls incur fees

What Tencent said

Tencent (腾讯) moved to calm online debate around the viral open‑source agent OpenClaw (“小龙虾”), laying out security, cost and product‑positioning details after users raised alarms on social media. The company said OpenClaw is an open agent framework whose safety depends on how and where users deploy it. Local installs are supported — and recommended on spare machines rather than primary workstations — while cloud deployments are integrated with Tencent Cloud (腾讯云) offerings such as Lighthouse, ADP and cloud desktop to provide end‑to‑end isolation from private data. It has been reported that Tencent is also standardizing plugin sources, permission declarations and curated Skills through a SkillHub to reduce supply‑chain risk.

Costs, charges and the offline plan

Installation of OpenClaw itself is free, Tencent emphasized. So why did some users report unexpected bills? Reportedly a small number of users saw charges after participating in a free installation drive; Tencent says those amounts — in one widely shared case about RMB 200 — were historical model‑call fees unrelated to the one‑time install. Tencent reiterated that model inference calls (token usage) generate charges, a common industry practice, and has launched a Coding Plan subscription for developers that bundles access to models including Tencent Hunyuan, GLM, Kimi and MiniMax.

Product map and adoption

Internally, Tencent draws a distinction between two “lobster” lines: community OpenClaw packages wrapped for easier deployment, and Tencent’s own desktop agent WorkBuddy, which shares the same agent architecture as the AI code assistant CodeBuddy and is described as Tencent’s in‑house “small lobster.” WorkBuddy is compatible with OpenClaw Skills but adds usability, model‑selection and tightened security controls. It has been reported that more than 10,000 Tencent employees have already experimented with internal OpenClaw builds, and agent‑driven development is being piloted across several internal products.

Bigger picture

This clarification comes amid a global scramble over AI models, regulatory scrutiny and shifting trade policy that affect model availability and deployment choices for Chinese cloud vendors and their customers. Tencent’s messaging aims to reassure enterprise and consumer users that open‑source agents can be used safely while making clear that the economics of calling large models will follow market norms — and that occasional offline free‑installation activities are being planned to broaden access.