# LLaVA-VL/LLaVA-NeXT

Repository: https://github.com/LLaVA-VL/LLaVA-NeXT
Canonical: https://ross.abutalabs.com/products/llava-next
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-06-15T06:32:49+00:00

## Health v2 (maintenance only)
Score: 64/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 87, release rhythm 35, longevity 64
- inputs: {"age_days": 908, "days_push": 79, "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 4716, forks 472 (observed 2026-08-28T04:08:57.433819+00:00)

## What it is
LLaVA-NeXT is a collection of open large multimodal models (LLaVA-NeXT, LLaVA-Video, LLaVA-OneVision, LLaVA-Critic-R1) that combine vision encoders with language models for image, video, and interleaved multimodal understanding. The repository provides model checkpoints, inference code, and a legacy training pipeline, with newer training moved to the lmms-engine project.

## Use cases
- run a vision-language model on images and videos
- train or fine-tune a multimodal LLM
- build an image and video question-answering chatbot
- evaluate open multimodal models
- process interleaved image-text inputs with an LLM
- download open VLM checkpoints for research

## When to choose
- you need open-weight multimodal models for image, video, or interleaved understanding
- you want to fine-tune or study LLaVA-family VLMs
- you need a strong open 7B-scale VLM with critic/reasoning variants

## When to avoid
- you need the latest training pipeline - use lmms-engine instead, as this repo's training code is legacy
- you only need closed-source frontier multimodal APIs
- you need production inference serving rather than research code

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, llm-training, image-processing, video-processing
- domain: large-language-models, computer-vision, deep-learning, artificial-intelligence
- platform: python
- tags: multimodal, vision-language-model, vlm, llava, model-training, open-source-models, gpu, linux

## Member repositories
- LLaVA-VL/LLaVA-NeXT (main) score 64

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:57.433819+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 2026-08-29T18:19:07.474457+00:00, confidence not recorded.
  - readme: https://github.com/LLaVA-VL/LLaVA-NeXT (fetched 2026-08-28T04:08:57.433819+00:00, sha 9806d8f20056)
- Data as of 2026-08-30T08:39:29.467469+00:00.