GLM-5.2 beats GPT-5.5 on long coding tasks while staying open
zhipu ai released glm-5.2, an open-source model scoring 74.4% on frontierswe, beating gpt-5.5 and nearly matching claude opus 4.8 in long-context coding tasks.
zhipu ai released glm-5.2 under the mit license, featuring a 1-million-token context window specifically architected for multi-hour coding agent workflows. the model scored 74.4% on the frontierswe benchmark, beating gpt-5.5 and landing just 1.4 percentage points behind claude opus 4.8, while also outperforming competitors on the posttrainbench. performance on general reasoning remains lower than proprietary leaders, though the model leads on aime 2026 math. independent analysis by artificial analysis ranks it as the strongest open-weights model currently available.