Z.ai debuts GLM-5.2 with stable 1M-token context and top coding scores (opens in new tab)
The update delivers leading open-source scores on multiple coding benchmarks, including 81.0 on Terminal-Bench 2.1 and 62.1 on SWE-bench Pro. Key technical innovations such as IndexShare and improved speculative decoding boost efficiency and enable practical large-scale usage. GLM-5.2 is available under an MIT open-source license with no regional limits. |
Read the original article