Jina AI: Jina VLM

jina-vlm is a 2.4B parameter vision-language model that achieves state-of-the-art multilingual visual question answering among open 2B-scale VLMs. The model couples a SigLIP2-So400M vision encoder (4…

jina/jina-vlm

jina-vlm is a 2.4B parameter vision-language model that achieves state-of-the-art multilingual visual question answering among open 2B-scale VLMs. The model couples a SigLIP2-So400M vision encoder (449M parameters) with a Qwen3-1.7B language backbone through an attention-pooling connector that reduces visual tokens by 4× while preserving spatial information. Using overlapping image tiling with 12 tiles plus a global thumbnail, it processes images of arbitrary resolution up to 4K. Training data comprises approximately 5M multimodal samples and 12B text tokens across 29 languages, with roughly half in English and the remainder spanning high- and moderate-resource languages including Chinese, Arabic, German, Spanish, French, Italian, Japanese, Korean, and more.

Specifications & capabilities

Input
text, image
Output
text
Context window · tokens
32768
Published model ID
jina-vlm
Capabilities
multimodal, visionInput

Provider specifications & access

1–1 / 1

jina ↗

source model id
jina-ai/jina-vlm
hosting
jina-catalog-only
platforms
  • Jina AI
context length
32768
architecture
input_modalities
  • text
  • image
output_modalities
  • text
capabilities

    Source specification observations

    1–1 / 1

    jina ↗

    field
    context_length
    value
    32768
    normalized value
    32768
    scope
    source-model
    source id
    jina
    confidence
    1

    Original sources

    1–5 / 5

    Jina AI · Model page ↗

    Hugging Face · jinaai/jina-vlm ↗

    arXiv · Paper · 2512.04032 ↗

    Jina AI · Source snapshot ↗

    Jina AI · Catalog ↗

    Source links

    1–2 / 2

    Explore more models

    Models

    Loading
    EVIDENCE / COMPARISON

    Benchmarks

    Explore every benchmark

    Loading

    These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

    Install TensorTrace