Term
GLM-4.6
GLM-4.6 is Zhipu AI's open-weight MoE model from September 2025 — 355B parameters (32B active), 200k context, MIT-licensed, with a focus on coding performance.
GLM-4.6 — explained in more detail
Zhipu AI released GLM-4.6 on September 30, 2025 as an open-weight model under the MIT license. The Mixture-of-Experts architecture totals 355 billion parameters, of which only about 32 billion are active per forward pass — markedly reducing inference cost versus dense models of comparable size. The context window is 200,000 tokens.
Example / Practical use
On LiveCodeBench v6, GLM-4.6 takes first place at 82.8%, as well as on Humanity’s Last Exam (HLE). On AIME 2025 the model reaches 93.9% (third place), and on Terminal-Bench 40.5%. In practical coding comparisons, human evaluators rated GLM-4.6 about 48.6% as good as Claude Sonnet 4. The parallel vision variant GLM-4.6V (106B) ships native function calling for multimodal tools; a 9B variant (GLM-4.6V-Flash) is tuned for local deployment. In February 2026, GLM-5 succeeded the model as Z.ai’s flagship.
Entdecke mehr
Three Days with Claude Fable 5 — and Why Opus Stays My Daily Driver
Fable 5 fixed a bug in one shot that Opus failed at twice. I'm still not switching. My honest impressions from the trenches after three days.
GlossarGLM-5.1
GLM-5.1 is Z.ai's open-source flagship from April 2026 — 744B parameters (40B active), 200k context, designed for agentic coding with up to 8 hours of autonomous runtime.
LexikonComparing AI models — who builds what and how to choose
The major model families in 2026 at a glance. Who builds Claude, GPT, Gemini, Llama, Mistral, DeepSeek, Qwen — and which model to pick when.