Zhipu AI Unveils GLM-4.6: A Giant Leap in Large Language Models
Zhipu AI has unveiled GLM-4.6, the latest iteration in its GLM series. This new version focuses on agentic workflows and long-context reasoning, boasting 355B parameters and supporting local serving with vLLM and SGLang.
GLM-4.6 shows significant improvements over its predecessor, GLM-4.5, demonstrating clear gains across eight public benchmarks. It also performs on par with Claude Sonnet 4/4.6 on several tasks. The model's input window stretches to 200K tokens, with a maximum output of 128K tokens.
In terms of licensing and availability, GLM-4.6 is released under the MIT license. Its weights are accessible on both Hugging Face and ModelScope, enabling open deployment. Additionally, the model can be integrated with popular coding agents and is available via Z.ai API and OpenRouter. Zhipu AI has also targeted lower token consumption in applied tasks, with GLM-4.6 using 15% fewer tokens than GLM-4.5 on the extended CC-Bench, while approaching parity with Claude Sonnet 4.
GLM-4.6, with its enhanced capabilities and open deployment options, is set to advance the field of large language models. Its improved performance and efficiency make it a compelling choice for a variety of applications.
Read also:
- Hematology specialist and anemia treatment: The role of a hematologist in managing anemia conditions
- Trump announces Chinese leader's confirmation of TikTok agreement
- U.S. Army Europe & Africa Bolsters NATO, African Partnerships in Phase Zero
- Hackers Utilize GOLD SALEM to Infiltrate Networks and Evade Security Measures, Deploying Warlock Ransomware
