# zai-org/CogVLM

a state-of-the-art-level open visual language model | 多模态预训练模型

Repository: https://github.com/zai-org/CogVLM
Canonical: https://ross.abutalabs.com/products/cogvlm
Language: Python
License: Apache-2.0
License Family: permissive
Topics: cross-modality, language-model, multi-modal, pretrained-models, visual-language-models
Last push: 2024-05-29T10:01:33+00:00

## Health v2 (maintenance only)
Score: 28/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 77
- inputs: {"age_days": 1081, "days_push": 826, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6744, forks 451 (observed 2026-08-28T04:09:48.239663+00:00)

## What it is
CogVLM is an open-source visual language model (17B) combining a vision encoder with a pretrained language model for image understanding and multi-turn dialogue, with CogAgent (18B) extending it for high-resolution GUI agent tasks. The repo provides model checkpoints, CLI and web demo inference, OpenAI Vision-compatible serving, and finetuning code.

## Use cases
- answer questions about images with a vision-language model
- generate captions for images
- build a GUI automation agent that understands screenshots
- finetune a multimodal VLM on custom data
- serve an OpenAI Vision-compatible image chat API
- run visual question answering on documents and charts

## When to choose
- you need an open-source VLM for image understanding or VQA
- you want to build a GUI agent that operates from screenshots
- you need a self-hosted multimodal model with finetuning support

## When to avoid
- you lack GPU hardware (multi-billion-parameter models need significant VRAM)
- you need the newest multimodal models rather than a 2024-era release
- you want a lightweight CPU-only image captioning solution

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, llm-inference, computer-vision, nlp
- domain: artificial-intelligence, large-language-models, computer-vision
- platform: python
- tags: visual-language-model, multimodal, vlm, image-understanding, gui-agent, cogagent, pretrained-model, vqa, image-captioning, natural-language-processing, gpu, linux

## Member repositories
- zai-org/CogVLM (main) score 28

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:48.239663+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:42:39.202000+00:00, confidence not recorded.
  - readme: https://github.com/zai-org/CogVLM (fetched 2026-08-28T04:09:48.239663+00:00, sha eaf5b3c4e3e7)
- Data as of 2026-08-30T08:39:29.467469+00:00.