GLM-5.1
Zhipu AI·
LLMsopen-weightFrontier
Overview
744B parameter MoE model, 40B active per forward pass. MIT license. Trained on Huawei Ascend hardware.
Capabilities and innovations
744B total / 40B active MoETrained on Huawei AscendMIT licenseAscend-native training pipelineEfficient MoE inference
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| SWE-bench Pro | 58.4% | unsourced |
Architecture and hardware
- Parameters
- 744B total, 40B active per token (MoE)
- Estimated VRAM at Q4
- ~428 GB, Frontier class
- Quantization formats
- GGUF
- Recommended runtime
- Ollama
- License
- MIT
Links
More from Zhipu AI
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.