GLM 4.5 Air (Thinking): benchmark results

Provider: Zhipu. Released 2025-07-28. Access: Open.

Unified ELO 1544 ± 1, rank #1021 of 3078 rated models, from 146 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval Lithuanian NLU - WikiANN LT70.41Named entity recognition Score (%)96.1
EuroEval Latvian NLU - Fullstack NER LV71.41Named entity recognition Score (%)95.7
EuroEval English Knowledge94.32Knowledge Average Score (%)95
EuroEval Icelandic NLU - MIM-GOLD NER75.93Named entity recognition Score (%)94.4
EuroEval Italian NLU - MultiNERD IT82.52Named entity recognition Score (%)93.2
EuroEval Dutch NLU - CoNLL NL72.82Named entity recognition Score (%)92.4
EuroEval Norwegian NLU - NorNE NN79.48Named entity recognition Score (%)91.9
EuroEval Portuguese NLU - HAREM58.3Named entity recognition Score (%)91.9
EuroEval German NLU - Sb10K58.92Sentiment classification Score (%)91.7
EuroEval French Knowledge77.58Knowledge Average Score (%)91.5
EuroEval Spanish NLU - CoNLL ES76.98Named entity recognition Score (%)91.5
EuroEval Icelandic Knowledge44.5Knowledge Average Score (%)91.4

Interactive version: theaggregate.ai/model?slug=glm-4-5-air-thinking · How It Works · Data refreshed daily, snapshot 2026-09-19.