GLM-OCR
Sovereignty and hosting
This model is catalogued for reference. It is not hosted on LLM EU yet, and no first-party endpoint serves it here. Requests to it go to the provider directly, under that provider's own terms.
Summary
GLM-OCR is a small multimodal model for document text extraction, built from a CogViT vision encoder and a GLM-0.5B language decoder. The vendor states it has about 0.9B parameters while the published safetensors total is 1.33B, and the configuration allows 131,072 text positions. It is MIT licensed and runs under vLLM, SGLang or Ollama for high-volume extraction.
Specifications
- Provider
- Z.ai (Zhipu AI) (CN)
- Context window
- 131k
- Parameters
- 1.33B
- Modalities
- vision
- License
- mit · licence
- Open weights
- yes
- Release date
- 30 Jan 2026
- Tasks
- vision, rag, legal-draft, cheap, summarization
Languages
Strong: zh, en
Adequate: de, fr, es, ru, ja, ko
Languages
Strong
- zh
- en
Adequate
- de
- fr
- es
- ru
- ja
- ko
Not verified: the provider has not documented this claim. — a language is listed as strong only where a source claims it; the card links the evidence where one exists.
When to use it
Strengths
- purpose-built document OCR rather than general vision
- runs on modest hardware and through common inference servers
- MIT licence
Limits
- text extraction only; it is not a general chat or reasoning model
- vendor parameter count (0.9B) differs from the published safetensors total (1.33B)
- no OCR benchmark has been run by LLM EU