All tasks / Models
GLM-OCR
Z.ai's open-weights ~1B multimodal OCR model (MIT) for parsing document text, tables and formulas.
- Z.ai
- Open weights
- Released 30 Jan 2026
Where it ranks
- SDocuments & OCRModels · Local · 95 on OmniDocBench v1.6 (end-to-end, v1.6_full)
Facts
- Vendor
- Z.ai
- Family
- GLM-OCR
- Released
- 30 Jan 2026
- Weights
- Open: download and run it yourself
- License
- MIT
- Parameters
- 1.33B
- GPU memory to run locally
- ~5 GB (Estimate, no official figure: BF16 weights ≈ 1.33 B × 2 GB + ~1.5 GB runtime overhead. Size is the Hugging Face safetensors total; OmniDocBench lists 0.9B.)
- Access
- Free (open weights)
- Last verified
- 18 Oct 2026
Price from huggingface.co, 18 Oct 2026.
Benchmark results
| Benchmark | Result | Entry on the leaderboard | As of | Source |
|---|---|---|---|---|
| IDP Leaderboard (v1.5) | 64.2% | GLM-OCROverall (mean of OlmOCR, OmniDoc and IDP Core columns) on IDP Leaderboard v1.5; page states its ranking is 'As of April 2026'. | 18 Oct 2026 | idp-leaderboard.org |
| MDPBench (multilingual document parsing) | 67.3% | GLM-OCROverall (public split, page-level mean of text, formula and table scores) from the board's leaderboard.json (updated 2026-10-04). | 4 Oct 2026 | huggingface.co |
| OCR Arena | 1,401 | GLM-OCRElo read from the site's JSON endpoint https://www.ocrarena.ai/api/leaderboard; 399 battles. | 18 Oct 2026 | ocrarena.ai |
| OmniDocBench v1.6 (end-to-end, v1.6_full) | 95 | GLM-OCROverall score in the README's end-to-end table 'OmniDocBench (v1.6_full)' (last leaderboard update 2026-09-11). | 11 Sep 2026 | github.com |
| ParseBench (models) | 29.6% | GLM-OCROverall (mean of tables, charts, content faithfulness, semantic formatting, visual grounding) from leaderboard.csv in the ParseBench repo (last leaderboard commit 2026-10-09). | 9 Oct 2026 | github.com |
Where you can use it
No tracked tool lists this model yet; as open weights it runs in local runtimes.