selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

Alternatives to ZwZ2-4B

120 open-weight models that do the same job as ZwZ2-4B (4.5B vision), ranked by how close they are in task and size, then by popularity and how recent they are. Licence and VRAM are shown for each.

the original
inclusionAI logo

ZwZ2-4B

inclusionAI·Vision

≈4.2 GB

ZwZ2-4B is the Qwen3.5-4B-based model in Zooming-without-Zooming-v2 (ZwZ2). It learns to connect visual evidence across regions through comparison, aggregation, matching, and multi-step reasoning. Training uses full images; inference requires no zooming tools.

4.5B256K ctx Unknown licence
1.1KOct 2026gguf
--vram
any
--gpus
--quant
  1. naver-hyperclovax logo

    HyperCLOVAX-SEED-Vision-Instruct-3B

    naver-hyperclovax·Vision

    ≈3.7 GB

    HyperCLOVAX-SEED-Vision-Instruct-3B is a model developed by NAVER, built upon its proprietary backbone model and fine-tuned through post-training. It is capable of understanding both text and images, as well as generating text.

    3.7B128K ctx hyperclovax-seed
    2.9KApr 2025gguf
    • Same category
    • Similar size (3.7B vs 4.5B)
    vs
  2. openbmb logo

    MiniCPM-V-4_5

    openbmb·Vision

    ≈6.8 GB

    A GPT-4o Level MLLM for Single Image, Multi Image and High-FPS Video Understanding on Your Phone

    8.7B40K ctx Apache 2.0
    67.8KAug 2025ollama
    • Same category
    • Similar size (8.7B vs 4.5B)
    • Commercial use OK
    vs
  3. google logo

    medgemma-4b-pt

    google·Vision

    ≈4.1 GB

    This section describes the MedGemma model and how to use it.

    4.3B128K ctx health-ai-developer-foundations
    558May 2025gguf
    • Same category
    • Similar size (4.3B vs 4.5B)
    vs
  4. MBZUAI logo

    AIN

    MBZUAI·Vision

    ≈5.8 GB

    Abstract Amid the swift progress of large language models (LLMs) and their evolution into large multimodal models (LMMs), significant strides have been made in high-resource languages such as English and Chinese.

    8.3B32K ctx MIT
    241.5KJan 2025gguf
    • Same category
    • Similar size (8.3B vs 4.5B)
    • Commercial use OK
    vs
  5. LiquidAI logo

    d1-2B

    LiquidAI·Vision

    ≈2.0 GB

    d1-2B is a decision model for text, pictures, and video, built on LFM3-VL-2B. It returns calibrated answers in one forward pass, without generating tokens.

    1.9B125K ctx lfm1.0
    17.1KOct 2026ollama
    • Same category
    • Smaller (1.9B vs 4.5B)
    • Newer release
    vs
  6. sensenova logo

    SenseNova-Loop-9B-Preview

    sensenova·Vision

    ≈7.2 GB

    SenseNova-Loop-9B-Preview is a multimodal language model developed for long-context and agentic workflows. It accepts text and images and generates text for applications that combine document understanding, coding, and interaction with external tools.

    9.7B256K ctx Apache 2.0
    5.9KOct 2026gguf
    • Same category
    • Larger (9.7B vs 4.5B)
    • Commercial use OK
    • Newer release
    vs
  7. Qwen logo≈5.8 GB

    We're excited to unveil Qwen2-VL, the latest iteration of our Qwen-VL model, representing nearly a year of innovation.

    8.3B32K ctx Apache 2.0
    551.4KAug 2024gguf
    • Same category
    • Similar size (8.3B vs 4.5B)
    • Commercial use OK
    vs
  8. LiquidAI logo

    LFM2-VL-3B

    LiquidAI·Vision

    ≈2.8 GB

    LFM2-VL-3B is the newest and most capable model in Liquid AI's multimodal LFM2-VL series, designed to process text and images with variable resolutions. Built on the LFM2 backbone, it extends the architecture for higher-capacity reasoning and stronger visual understanding while retaining efficiency.

    3B125K ctx lfm1.0
    3.6KOct 2025gguf
    • Same category
    • Similar size (3B vs 4.5B)
    vs
  9. moondream logo

    moondream3.1-9B-A2B

    moondream·Vision

    ≈6.5 GB

    Moondream 3.1 is a vision language model with a mixture-of-experts architecture (9B total parameters, 2B active). It delivers state-of-the-art visual reasoning and detection while staying fast and cheap to deploy.

    9.3BA2Bmoondream-model-license-1.0
    8.2KJun 2026
    • Same category
    • Larger (9.3B vs 4.5B)
    vs
  10. llm-jp logo

    llm-jp-4-vl-9b

    llm-jp·Vision

    ≈6.9 GB

    Compared with the earlier LLM-jp-4-VL 9B beta, the new model is trained on datasets that have no license or terms-of-use issues for commercial use. It is also enhanced to generate a reasoning trace before a final answer.

    9.1B64K ctx Apache 2.0
    4.3KAug 2026gguf
    • Same category
    • Similar size (9.1B vs 4.5B)
    • Commercial use OK
    vs
  11. google logo

    paligemma-3b-pt-224

    google·Vision

    ≈2.4 GB

    Transformers PaliGemma 3B weights, pre-trained with 224*224 input images and 128 token input/output text sequences. The models are available in float32, bfloat16 and float16 formats for fine-tuning.

    2.9BGemma Terms of Use
    226.8KMay 2024
    • Same category
    • Similar size (2.9B vs 4.5B)
    vs
  12. microsoft logo

    AesCode-8B

    microsoft·Vision

    ≈6.8 GB

    AesCode generates information-rich visual artifacts such as slides, posters, and dashboards as HTML/CSS. The output remains structured, editable, and verifiable, but the task poses a distinct challenge: code models cannot see how layout, hierarchy, and color come together on the canvas.

    8.8B256K ctx Apache 2.0
    1.9KSept 2026gguf
    • Same category
    • Similar size (8.8B vs 4.5B)
    • Commercial use OK
    vs
  13. fancyfeast logo≈6.5 GB

    JoyCaption is an image captioning Visual Language Model (VLM) built from the ground up as a free, open, and uncensored model for the community to use in training Diffusion models.

    8.5B128K ctx Unknown licence
    63.1KMay 2025gguf
    • Same category
    • Similar size (8.5B vs 4.5B)
    vs
  14. Tongyi-MAI logo

    MAI-UI-8B

    Tongyi-MAI·Vision

    ≈6.8 GB

    The development of GUI agents could revolutionize the next generation of human-computer interaction. Motivated by this vision, we present MAI-UI, a family of foundation GUI agents spanning the full spectrum of sizes, including 2B, 8B, 32B, and 235B-A22B variants.

    8.8B256K ctx Apache 2.0
    14.5KDec 2025gguf
    • Same category
    • Similar size (8.8B vs 4.5B)
    • Commercial use OK
    vs
  15. llava-hf logo

    llava-1.5-7b-hf

    llava-hf·Vision

    ≈5.2 GB

    Below is the model card of Llava model 7b, which is copied from the original Llava model card that you can find here.

    7.1B4K ctx Llama 2 Community License
    637.9KDec 2023gguf
    • Same category
    • Similar size (7.1B vs 4.5B)
    vs
  16. OpenGVLab logo

    InternVL3-8B

    OpenGVLab·Vision

    ≈5.7 GB

    We introduce InternVL3, an advanced multimodal large language model (MLLM) series that demonstrates superior overall performance.

    7.9B32K ctx Apache 2.0
    36.8KApr 2025gguf
    • Same category
    • Similar size (7.9B vs 4.5B)
    • Commercial use OK
    vs
  17. Qwen logo≈2.0 GB

    We're excited to unveil Qwen2-VL, the latest iteration of our Qwen-VL model, representing nearly a year of innovation.

    2.2B32K ctx Apache 2.0
    916.8KAug 2024gguf
    • Same category
    • Smaller (2.2B vs 4.5B)
    • Commercial use OK
    vs
  18. zai-org logo

    GLM-4.6V-Flash

    zai-org·Vision

    ≈6.9 GB

    This model is part of the GLM-V family of models, introduced in the paper GLM-4.1V-Thinking and GLM-4.5V: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning.

    10.3B128K ctx MIT
    62.7KDec 2025gguf
    • Same category
    • Larger (10.3B vs 4.5B)
    • Commercial use OK
    vs
  19. Qwen logo≈2.6 GB

    Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

    2.1B256K ctx Apache 2.0
    46.3KOct 2025gguf
    • Same category
    • Smaller (2.1B vs 4.5B)
    • Commercial use OK
    vs
  20. sensenova logo

    SenseNova-Loop-9B

    sensenova·Vision

    ≈7.2 GB

    🔥 2026-11-17: We release SenseNova-Loop-9B, with model weights and code available to the community.

    9.4B512K ctx Apache 2.0
    1.7KNov 2026gguf
    • Same category
    • Larger (9.4B vs 4.5B)
    • Commercial use OK
    • Newer release
    vs
  21. allenai logo

    MolmoPoint-8B

    allenai·Vision

    ≈6.8 GB

    MolmoPoint-8B is a fully-open VLM developed by the Allen Institute for AI (Ai2) that support image, video and multi-image understanding and grounding. It has new pointing mechansim that improves image pointing, video pointing, and video tracking, see our technical report for details.

    8.7B37K ctx Apache 2.0
    5.1KMar 2026
    • Same category
    • Similar size (8.7B vs 4.5B)
    • Commercial use OK
    vs
  22. google logo

    t5gemma-2-4b-4b

    google·Vision

    ≈6.8 GB

    T5Gemma is a family of lightweight yet powerful encoder-decoder research models from Google, built by adapting pretrained decoder-only Gemma models into encoder-decoder ones.

    8.9B128K ctx Gemma Terms of Use
    16.9KOct 2025
    • Same category
    • Similar size (8.9B vs 4.5B)
    vs
  23. moondream logo

    moondream3-preview

    moondream·Vision

    ≈7.1 GB

    Moondream 3 (Preview) is an vision language model with a mixture-of-experts architecture (9B total parameters, 2B active). This model makes no compromises, delivering state-of-the-art visual reasoning while still retaining our efficient and deployment-friendly ethos.

    9.3BCustom licence
    37.3KSept 2025
    • Same category
    • Larger (9.3B vs 4.5B)
    vs
  24. nvidia logo

    DAM-3B

    nvidia·Vision

    ≈2.9 GB

    Describe Anything Model 3B (DAM-3B) takes inputs of user-specified regions in the form of points/boxes/scribbles/masks within images, and generates detailed localized descriptions of images.

    3Bnvidia-non-commercial-license
    6.9KApr 2025
    • Same category
    • Similar size (3B vs 4.5B)
    vs