GLM-5.3 is Z.ai open-weight model aimed at coding and long-horizon agentic work. API pricing is $1.40 per million input tokens, $0.26 cached, and $4.40 output. Context is 1M tokens with 128K maximum output.
It uses the same base model as GLM-5.2 and is priced identically, with every gain coming from post-training. Z.ai reports roughly 50 per cent improvement on its internal Code Bench, Terminal-Bench 3.0 rising from 4.6 to 28.3, and DeepSWE v1.1 from 46.2 to 66.9.
Two limitations decide most evaluations. Input is text only, so screenshots, PDFs and charts are out of scope entirely. And reasoning is always on with no way to disable it - effort is low, high or max, defaulting to max - so simple calls still spend thinking tokens.
Weights are published on Hugging Face at 753B parameters in a mixture-of-experts configuration, so self-hosting is possible but needs serious hardware.