Chinese AI Model GLM-5.2 Surpasses Claude Opus 4.7 in Benchmarks
GLM-5.2 shows that Chinese open-source models continue to close the gap with top proprietary systems, matching or beating Anthropic's latest in specific coding and long-horizon tasks.
Reporting from 1 sources: GIGAZINE.
Chinese AI company Z.ai officially announced GLM-5.2 on June 17, 2026. The 753-billion-parameter model surpasses Claude Opus 4.7 in multiple benchmarks, outperforms Claude Fable 5 in human blind tests for coding, and is available as an open model under the MIT License.
Z.ai released GLM-5.2 on June 17, following a partial reveal on June 13 when Anthropic discontinued Claude Fable 5 and Claude Mythos 5 services. The model has 753 billion parameters and supports 1 million token input. In benchmark comparisons, GLM-5.2 surpassed GPT-5.5 in multiple tests and approached Claude Opus 4.8 scores. It excelled in long-duration task execution, beating both Claude Opus 4.7 and GPT-5.5. In the Arena coding performance test, GLM-5.2 ranked second worldwide, behind only Claude Opus 4.8. In the Design Arena coding test, it defeated Claude Fable 5 and took first place. The model is available on Z.ai's chat service and can be used with coding agents including ZCode, Claude Code, and OpenCode. It is distributed as an open model on Hugging Face under the MIT License.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.