Skip to content

Models · Vision

vikasit-vision-2b

Lightweight vision. Image captioning, OCR, visual Q&A on device.

Parameters
2B
Context window
32K
Best for
On-device vision

Benchmarks

MMMU (val)53.4%
MMMU-Pro36.5%
MathVista61.3%
DocVQA93.3%
AI2D76.9%

Thinking mode, no tools where applicable. See full comparisons on the benchmarks page.

Run it

# via Vikasit Inference (OpenAI-compatible)
base_url = "https://api.vikasit.ai"
model = "vikasit-vision-2b"