Skip to content

Commit d666f1a

Browse files
feat(gallery): add HunyuanOCR variants (#11540)
Add the official Q8 and BF16 llama.cpp builds for the archived HunyuanOCR 1.0 checkpoint. Assisted-by: Codex:gpt-5 Co-authored-by: localai-org-maint-bot <306269227+localai-org-maint-bot@users.noreply.github.com>
1 parent 1e3e72e commit d666f1a

1 file changed

Lines changed: 65 additions & 0 deletions

File tree

gallery/index.yaml

Lines changed: 65 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -13285,6 +13285,71 @@
1328513285
- filename: mmproj-GLM-OCR-Q8_0.gguf
1328613286
sha256: 9c4b58e33e316ed142eb5dcb41abec3844d3e6e5dc361ffb782c3fa9d175141f
1328713287
uri: huggingface://ggml-org/GLM-OCR-GGUF/mmproj-GLM-OCR-Q8_0.gguf
13288+
- &hunyuan-ocr
13289+
name: hunyuan-ocr-q8
13290+
variants:
13291+
- model: hunyuan-ocr-bf16
13292+
url: github:mudler/LocalAI/gallery/virtual.yaml@master
13293+
urls:
13294+
- https://huggingface.co/tencent/HunyuanOCR/tree/main/v1.0
13295+
- https://huggingface.co/ggml-org/HunyuanOCR-GGUF
13296+
description: |
13297+
HunyuanOCR is Tencent's lightweight vision-language model for document parsing, text spotting, information extraction, and text-image translation. This Q8_0 GGUF build runs on llama.cpp with its bundled vision projector.
13298+
license: tencent-hunyuan-community
13299+
icon: https://huggingface.co/tencent.png
13300+
tags:
13301+
- llm
13302+
- gguf
13303+
- gpu
13304+
- image-to-text
13305+
- ocr
13306+
- multimodal
13307+
- cpu
13308+
- hunyuan
13309+
last_checked: "2026-08-15"
13310+
overrides:
13311+
backend: llama-cpp
13312+
known_usecases:
13313+
- chat
13314+
- vision
13315+
mmproj: mmproj-HunyuanOCR-Q8_0.gguf
13316+
options:
13317+
- use_jinja:true
13318+
parameters:
13319+
model: HunyuanOCR-Q8_0.gguf
13320+
template:
13321+
use_tokenizer_template: true
13322+
files:
13323+
- filename: HunyuanOCR-Q8_0.gguf
13324+
sha256: cdafc794cafeae377868d7a40a70e282a737e39abe77c0d8b73614447b364a21
13325+
uri: huggingface://ggml-org/HunyuanOCR-GGUF/HunyuanOCR-Q8_0.gguf
13326+
- filename: mmproj-HunyuanOCR-Q8_0.gguf
13327+
sha256: b77913164ff73d4c0dc4d994e236ed72bacbbe5c5db1ec9b2828627b46c32804
13328+
uri: huggingface://ggml-org/HunyuanOCR-GGUF/mmproj-HunyuanOCR-Q8_0.gguf
13329+
- !!merge <<: *hunyuan-ocr
13330+
name: hunyuan-ocr-bf16
13331+
description: |
13332+
HunyuanOCR in BF16 GGUF format for maximum model and vision-projector fidelity. It runs on llama.cpp and supports document parsing, text spotting, information extraction, and text-image translation.
13333+
overrides:
13334+
backend: llama-cpp
13335+
known_usecases:
13336+
- chat
13337+
- vision
13338+
mmproj: mmproj-HunyuanOCR-bf16.gguf
13339+
options:
13340+
- use_jinja:true
13341+
parameters:
13342+
model: HunyuanOCR-bf16.gguf
13343+
template:
13344+
use_tokenizer_template: true
13345+
files:
13346+
- filename: HunyuanOCR-bf16.gguf
13347+
sha256: a160215620dbd0ab43ec6faa28259654fd24c929953aa97c765176f7c0363217
13348+
uri: huggingface://ggml-org/HunyuanOCR-GGUF/HunyuanOCR-bf16.gguf
13349+
- filename: mmproj-HunyuanOCR-bf16.gguf
13350+
sha256: 46401739a91d0778d86369bb952db685b215512d61a941c3b859f337f6014fcd
13351+
uri: huggingface://ggml-org/HunyuanOCR-GGUF/mmproj-HunyuanOCR-bf16.gguf
13352+
variants: []
1328813353
- name: deepseek-ocr
1328913354
url: github:mudler/LocalAI/gallery/virtual.yaml@master
1329013355
urls:

0 commit comments

Comments
 (0)