GLM-Z1-AirX
Zhipu's ultra-fast reasoning model hitting 200 tokens per second.
About
GLM-Z1-AirX was released by Zhipu AI on April 14, 2025, as the speed tier of its GLM-Z1 reasoning model family, alongside GLM-Z1-Air and the free GLM-Z1-Flash. Zhipu claimed it as the fastest inference model in China at the time, reaching up to 200 tokens per second, about eight times faster than conventional models.
It is an API-only hosted model served through Zhipu's bigmodel.cn platform, aimed at real-time interaction and high-throughput request scenarios. The companion open-source GLM-Z1-32B-0414 and 9B models cover the self-hosted side of the family.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 101; Trending score 78/100 from researcher assessment.