Z AI
GLM 4.5 Air
GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications. Like GLM-4.5, it adopts the Mixture-of-Experts (MoE) architecture but with a more compact parameter size. GLM-4.5-Air also supports hybrid inference modes, offering a "thinking mode" for advanced reasoning and tool use, and a "non-thinking mode" for real-time interaction. Users can control the reasoning behaviour with the `reasoning` `enabled` boolean. [Learn more in our docs](https://openrouter.ai/docs/use-cases/reasoning-tokens#enable-reasoning-with-default-config)
- Input / 1M tokens
- $0.130
- Output / 1M tokens
- $0.850
- Context window
- 131K tokens
- Provider
- Z AI
- Cached input / 1M
- $0.025
- Knowledge cutoff
- 2024-12-31
Performance
Median streaming throughput and first-token latency measured by Artificial Analysis.
- Output tokens / sec
- 67 t/s
- Time to first token
- 1.25s
Benchmarks
Intelligence, coding, and math indexes plus the underlying evaluation scores.
- Intelligence Index
- 23
- Coding Index
- 24
- Math Index
- 81
- MMLU-Pro
- 81.5%
- GPQA
- 73.3%
- HLE
- 6.8%
- LiveCodeBench
- 68.4%
- SciCode
- 30.6%
- MATH-500
- 96.5%
- AIME
- 67.3%
Benchmarks via Artificial Analysis