GLM-5.2 Now Processes 40% of Developer Tokens on OpenRouter, Undercutting Opus 4.8 by Up to 82%

GLM-5.2, Z.ai's open-weight coding model, has moved from topping benchmark leaderboards to dominating real developer usage: it now accounts for roughly 40% of coding tokens processed through OpenRouter, according to a new analysis of platform traffic.

• Z.ai prices GLM-5.2 at $1.40/$4.40 per million input/output tokens, versus $5/$25 for Claude Opus 4.8, a 72% and 82% discount respectively
• Some OpenRouter-listed providers have pushed GLM-5.2 input pricing as low as $0.91 per million tokens
• Production teams report effective costs running at 20-35% of Opus 4.8 once retries and prompt caching are factored in, a smaller gap than the sticker prices imply but still substantial
• The shift tracks a broader trend: Chinese-developed models overtook US providers in weekly OpenRouter token volume for the first time in February 2026

The token share follows GLM-5.2's chart-topping result on a one-shot coding benchmark suite in June, suggesting its eval strength is carrying over into production agentic workloads. Compare GLM-5.2 yourself.