ZCode Drops 300M Free Tokens for GLM-5.3-Flash Over the Weekend
A 300-million-token trial of a Claude Opus-class multimodal model at near-zero cost lets developers stress-test long-context reasoning, browser automation, and computer-use workflows over a full weekend without touching a paid plan.
A weekend token giveaway puts 300 million free tokens of GLM-5.3-Flash into every logged-in ZCode user's account. The model is the first natively multimodal entry in the GLM-5 series, supports a 1-million-token context window, and matches Claude Opus 4.8 on intelligence benchmarks at roughly one-fortieth the cost. Inside ZCode, it can be paired with Browser Use and Computer Use to drive a developer's local machine directly.
The credits become active Friday at 20:00 Beijing time and expire Monday at 09:00. Claims open three hours earlier, at 15:00 Friday, through a card in the client's lower-left corner. The pool is capped and first-come-first-served; each account gets one claim, and unused tokens simply vanish at the deadline. Paying users keep their existing plan quotas untouched and get priority if the service throttles under load.
A weekend-only burst of free high-tier tokens is a distribution play: it puts a Claude Opus-competitive model into developers' hands for long enough to form a habit, without cannibalizing paid plans.
Pairing the giveaway with Browser Use and Computer Use signals that ZCode is betting on agentic desktop workflows — not just chat — as the sticky use-case that converts trial users into subscribers.