zai-org/VisualGLM-6B

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

4154 stars · 419 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

VisualGLM-6B is an open-source multimodal conversational language model supporting images, Chinese, and English, built on ChatGLM-6B with a BLIP2-Qformer visual bridge (7.8B parameters total). It supports quantized deployment on consumer GPUs and offers both Hugging Face and SwissArmyTransformer interfaces.

Use cases

• chat with a model about an image
• generate image captions in Chinese or English
• run a multimodal LLM locally on a consumer GPU
• fine-tune a vision-language model with LoRA or P-tuning
• build a bilingual image question-answering bot

When to choose

• you need a bilingual Chinese/English multimodal chat model
• you have limited GPU memory and need INT4 quantized inference
• you want parameter-efficient fine-tuning of a vision-language model

When to avoid

• you need state-of-the-art multimodal performance (the project recommends CogVLM instead)
• you need an English-only model
• you cannot tolerate hallucinations or factual errors in image descriptions

Facets

library · maturity maintenance

llm-inference machine-learning chatbot image-processing large-language-models artificial-intelligence computer-vision python multimodal vision-language-model chatglm blip2 quantization chinese-english image-captioning natural-language-processing gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="zai-org/VisualGLM-6B")