ByteDance’s Volcano Engine, the cloud unit that released an OpenClaw-based cloud agent tool ArkClaw, is betting that the next phase of artificial intelligence will hinge on cheaper tokens, higher inference efficiency and longer context windows.

“Agent-related token consumption still accounts for a single-digit percentage of total token usage, but it is growing,” said Li Guodong, chief architect of ArkClaw, on Tuesday on the sidelines of OpenClaw’s first mainland China event since the open-source agent framework went viral globally earlier this year, triggering a wave of “lobster” enthusiasm among Chinese developers. The nickname lobster comes from the AI tool’s logo.

Although some of the initial hype around OpenClaw has eased, the Shanghai event was packed with about 1,300 attendees, according to the organiser, the Mu, a global builder community that was forced to restrict entry at one point. Attendees wore claw-themed headgear and gathered around demos and keynotes.

Volcano Engine began developing agent-relevant products last year and reached out to OpenClaw soon after the project gained traction during the Chinese New Year holiday, Li said.

In March, it launched ArkClaw, a cloud version of OpenClaw that Li likened to turning MySQL – a widely used open-source database – into a service. In April, the two sides co-launched the China mirror site for ClawHub, OpenClaw’s skills marketplace.

Token demand is already climbing amid rapid adoption and advances in multimodal models. As of March, daily average token usage of ByteDance’s Doubao large language models had surpassed 120 trillion, doubling in three months and rising more than 1,000-fold from their May 2024 launch, according to Volcano Engine.