GLM-5.2
Z.aiZ.ai's flagship open-weights frontier model (mid-June 2026, successor to GLM-5.1), built for long-horizon agentic and coding tasks. A sparse MoE (~753B total params) with a 1M-token context, introducing 'IndexShare' sparse attention (~2.9x FLOP cut at 1M context) and improved speculative decoding. Positioned as a leading open-weights coding model.
MIT-licensed open weights (HF model card + LICENSE, (c) Zhipu AI 2026). ~753B-param MoE, 1M context. Consolidated on 2026-07-29 from glm-4, glm-4-6, glm-4-9b, glm-5-1; openness follows glm-5-2, the release that currently governs. See docs/reference/identity.md. GLM-5.2 has pulled away sharply since July and now supplies most of the family's downloads, so the adoption band, which was set on the GLM-4 generation, is a level low and flagged for review. Verified 2026-08-13 via the GLM-5.2 LICENSE file, its HF model card and the zai-org/GLM-5 repository.
Openness
3 high confidence- weights
- open(MIT)
- data
- closed
- code
- open(inference)
- license
- MIT(OSI)
MIT weights on the current release, data closed, no training pipeline, so 3. Z.ai moved to MIT partway through the family: GLM-4 and GLM-4-9B carried the GLM-4 License, which requires commercial-use registration and a 'Built with GLM' attribution and scored 2, while GLM-4.6, GLM-5.1 and GLM-5.2 are plain MIT. GLM-5.2 governs as the current release. Capability 5 comes from the GLM-5 line; the GLM-4 line scored 2.
- https://huggingface.co/zai-org/GLM-5.2/blob/main/LICENSE recorded 2026-08-13
"MIT License / Copyright (c) 2026 Zhipu AI / Permission is hereby granted, free of charge ... without restriction" - the unmodified MIT text.
- https://huggingface.co/api/models/zai-org/GLM-5.2 recorded 2026-08-13
card license `mit`, `gated: false`, `private: false`, 2,662,328 downloads and 4,944 likes; no corpus declared on the card.
- https://api.github.com/repos/zai-org/GLM-5 recorded 2026-08-13
`private: false`, `archived: false`, license `Apache-2.0`, 6,943 stars, description "GLM-5: From Vibe Coding to Agentic Engineering". Inference and deployment code, published.
Adoption
4 high confidence2,774,206 downloads in the trailing 30 days across the four declared artifacts (zai-org/GLM-4.6 28,016; zai-org/glm-4-9b 11,688; zai-org/GLM-5.1 72,174; zai-org/GLM-5.2 2,662,328), which bands at level 4 (1M-10M) on the model adoption scale. GLM-5.2 alone supplies 96% of that total, so the current release carries the family rather than a generation-old checkpoint. The older THUDM/glm-4-9b repository now redirects to zai-org.
- https://huggingface.co/api/models/zai-org/GLM-4.6 recorded 2026-08-14
28,016 downloads in the trailing 30 days for zai-org/GLM-4.6; 1,230 likes
- https://huggingface.co/api/models/THUDM/glm-4-9b recorded 2026-08-14
11,688 downloads in the trailing 30 days for THUDM/glm-4-9b (now served as zai-org/glm-4-9b)
- https://huggingface.co/api/models/zai-org/GLM-5.1 recorded 2026-08-14
72,174 downloads in the trailing 30 days for zai-org/GLM-5.1; 1,839 likes
- https://huggingface.co/api/models/zai-org/GLM-5.2 recorded 2026-08-14
2,662,328 downloads in the trailing 30 days for zai-org/GLM-5.2; 4,948 likes; gated false
- https://huggingface.co/zai-org/GLM-4-9B-Chat recorded 2026-06-04
GLM-4-9B-Chat model card: 121,169 downloads last month; the undeclared, generation-behind SKU the old band was read on
Capability
5 high confidenceSWE-Bench Pro 58.4 (reported #1 at release, edging GPT-5.4 57.7 and Claude Opus 4.6 57.3); Terminal-Bench 2.0 63.5. Frontier-tier agentic coding for an open-weights model.
- https://serenitiesai.com/articles/glm-5-1-zhipu-coding-benchmark-claude-opus-comparison-2026 recorded 2026-08-13
SWE-Bench Pro 58.4 against GPT-5.4 57.7 and Claude Opus 4.6 57.3, and Terminal-Bench 2.0 63.5 - all four figures the band rests on
- https://www.modemguides.com/blogs/ai-news/glm-5-1-open-source-benchmarks-local-ai recorded 2026-08-13
SWE-Bench Pro 58.4 present; corroborates the serenitiesai reading. Does not carry the Terminal-Bench figure.
- https://www.vals.ai/models/zai_glm-5.1-thinking recorded 2026-08-13
The page returns HTTP 404 - it establishes nothing and the band does not rest on it.
Verified 2026-08-13