Project AURIONAI Use-Cases Research & Innovation
Z.ai (Zhipu)

Z.ai (Zhipu)

Company, China. 30 models in the directory.

companyChina
0deployments

Identity

Type
company
Country label
China
HQ country
China
Founding country
China
Parent
not in the notes
Parent country
not in the notes
Country basis
the research notes
Where this comes from
Convention D1: a dual label, HQ first, when the HQ, founding or parent country differ.

Models by Class

LLM18

Models

ModelClassOpennessParametersReleased
Z.ai (Zhipu)GLM-5.3LLMopen weights restricted753B25 Aug 2026Z.ai (Zhipu)GLM-5.3-Flashmultimodal LLMopen weights permissive320B25 Aug 2026Z.ai (Zhipu)GLM-5.2LLMopen weights permissive753B13 Jun 2026Z.ai (Zhipu)GLM-5.1LLMunknown–3 Apr 2026Z.ai (Zhipu)GLM-5V-Turbomultimodal LLMunknown–1 Apr 2026Z.ai (Zhipu)GLM-OCROCR and documentopen weights permissive900M30 Jan 2026Z.ai (Zhipu)GLM 4.7 FlashLLMunknown–19 Jan 2026Z.ai (Zhipu)GLM-Imageimage generationopen weights permissive–Jan 2026Z.ai (Zhipu)GLM-ASR-Nano (GLM-ASR-Nano-2512)speech recognitionopen weights permissive1.5B9 Dec 2025Z.ai (Zhipu)GLM-4.6Vmultimodal LLMopen weights permissive106B8 Dec 2025Z.ai (Zhipu)GLM-4.5Vmultimodal LLMopen weights permissive106B11 Aug 2025Z.ai (Zhipu)GLM-4.5LLMunknown355B8 Aug 2025Z.ai (Zhipu)CogVideoX-I2Vworld modelunknown–16 Sep 2024Z.ai (Zhipu)GLM-4-9B-ChatLLMunknown9.4B4 Jun 2024Z.ai (Zhipu)GLM-4V-9Bmultimodal LLMunknown13.91B4 Jun 2024Z.ai (Zhipu)CogAgent chatZ.ai (Zhipu), Tsinghua Universitymultimodal LLMopen weights permissive18.29B14 Dec 2023Z.ai (Zhipu)CogVLM chatZ.ai (Zhipu), Tsinghua Universitymultimodal LLMopen weights permissive17.64B16 Nov 2023Z.ai (Zhipu)ChatGLM3-6BZ.ai (Zhipu), Tsinghua UniversityLLMunknown6.24B25 Oct 2023Z.ai (Zhipu)CogVLMZ.ai (Zhipu), Tsinghua Universitymultimodal LLMopen weights permissive–5 Oct 2023Z.ai (Zhipu)ChatGLMZ.ai (Zhipu), Tsinghua UniversityLLMunknown–24 Jun 2023Z.ai (Zhipu)ChatGLM2-6BZ.ai (Zhipu), Tsinghua UniversityLLMunknown–24 Jun 2023Z.ai (Zhipu)GLMLLMunknown––Z.ai (Zhipu)GLM-4 (0520)LLMproprietary––Z.ai (Zhipu)GLM-4.5-Air-BaseLLMunknown––Z.ai (Zhipu)GLM-4.6LLMunknown––Z.ai (Zhipu)GLM-4.7LLMunknown––Z.ai (Zhipu)GLM-4-9B-0414LLMunknown––Z.ai (Zhipu)GLM-5LLMunknown––Z.ai (Zhipu)GLM-5.3-FlashXLLMproprietary––Z.ai (Zhipu)GLM-5.3-PrimeLLMproprietary––

Scores

BenchmarkModelScoreWho ran itSetting
Arena Text leaderboardGLM-5.11,464 Eloindependent
Arena Text leaderboardGLM-5.21,475 Eloindependentglm-5.2-max
Arena Text leaderboardGLM-5.31,478 Eloindependentglm-5.3-max
Arena Text leaderboardGLM-5.3-Flash1,475 Eloindependent
Artificial Analysis Intelligence IndexGLM-5.345 indexindependentGLM-5.3 (Max)
Artificial Analysis Intelligence IndexGLM-5.345 indexindependentmax
Artificial Analysis Text-to-Image ArenaGLM-Image891 Eloindependent
BFCLGLM-4 (0520)81.76 scorevendor-reportedBerkeley Function Call overall
DeepSWEGLM-5.366.9 %vendor-reportedmini-swe-agent
GPQAGLM-4 (0520)39.9 %vendor-reported
GSM8KGLM-4 (0520)93.3 %vendor-reported
HumanEvalGLM-4 (0520)78.5 %vendor-reported
Humanity's Last Exam (HLE)GLM-5.362.5 %vendor-reportedwith tools
IFEvalGLM-4 (0520)79.1 %vendor-reportedEnglish, prompt-level strict
MATHGLM-4 (0520)61.3 %vendor-reported
MMLUGLM-4 (0520)83.3 %vendor-reported
OmniDocBenchGLM-OCR95.22 scorevendor-reportedoverall
Real5-OmniDocBenchGLM-OCR90.32 scorevendor-reportedoverall
SEA-HELMGLM 4.7 Flash53.76 scoremaintainer-runCheckpoint GLM-4.7-Flash as listed; release view (data last updated 18 Sep 2026); score is the average of 2000 bootstraps
SEA-HELMGLM 4.7 Flash67.89 scoremaintainer-run
SWE-bench ProGLM-5.262.1 %vendor-reportedmax effort
SWE-bench VerifiedGLM-5.280 %independentthird-party via Thinking Machines
tau2-benchGLM-5.299.1 %independentArtificial Analysis telecom run
Terminal-BenchGLM-5.281 %vendor-reported
Terminal-BenchGLM-5.388.2 %vendor-reported
Terminal-BenchGLM-5.388.2 %vendor-reportedClaude Code 2.1.207
Terminal-BenchGLM-5.328.3vendor-reported
WorldScoreCogVideoX-I2V59.12 scoremaintainer-runDynamic aggregate
WorldScoreCogVideoX-I2V62.15 scoremaintainer-runStatic aggregate
Where this comes from
Every recorded score of this organisation's models, with who ran it. Scores from different settings are not directly comparable.

Sources

  1. Pricing - Overview - Z.AI DEVELOPER DOCUMENT docs.z.ai https://docs.z.ai/guides/overview/pricing
  2. huggingface.co/zai-org https://huggingface.co/zai-org
  3. huggingface.co/zai-org/GLM-5.3 https://huggingface.co/zai-org/GLM-5.3
  4. arxiv.org/pdf/2606.03264 https://arxiv.org/pdf/2606.03264
  5. huggingface.co/api/models/zai-org/GLM-OCR https://huggingface.co/api/models/zai-org/GLM-OCR
  6. τ²-bench results Leaderboard & Scores , October 2026 benchlm.ai https://www.benchlm.ai/benchmarks/tau2-bench
  7. Over 30 AI models have been trained at the scale of GPT-4 epoch.ai https://epoch.ai/data-insights/models-over-1e25-flop
  8. GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models arxiv.org https://arxiv.org/abs/2508.06471
  9. yicaiglobal.com/news/zhipu-ai-soars-in-hong-kong-stock-market... https://www.yicaiglobal.com/news/zhipu-ai-soars-in-hong-kong-stock-market-debut-as-chinese-startup-becomes-worlds-first-llm-firm-to-go-public

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Z.ai (Zhipu)", https://aurion.semai.id/organisations/zhipu-zai/, records of October 2026.