StopDown
News · 7 May 2026ArcleIntelligenceMultimodal AI model builder

Student builds AI model and seeks funding

By · Editor ·

  • Abhinav Anand built a multimodal AI model using personal savings and cloud credits
  • The model processes text, images, audio, and video with over two million tokens context
  • He seeks $35,000 to complete the pipeline and plans to open-source the code

What happened

  • Abhinav Anand is a 19-year-old student from Bihar who built an AI model independently.
  • He spent Rs 11 lakh of personal savings to develop the ArcleIntelligence project.
  • The model features a 5.82 billion parameter architecture for processing multiple media types.
  • Compute costs were kept low at approximately Rs 64,000 for GPU usage.
  • Anand is seeking $35,000 in funding to finish his development pipeline.
  • He plans to release weights on Hugging Face and code on GitHub.
  • Previous projects include a YouTube analytics app and a voice assistant system.
  • He has no computer science degree or team backing this specific venture.

Why it matters

  • The story highlights independent development by a high school student without formal training.
  • Funding sources included grants from RunPod and credits from DigitalOcean.
  • The project demonstrates accessible tools for building complex AI models.

The business

  • ArcleIntelligence handles text, image, document, audio, and video processing tasks.
  • Image generation supports a resolution of 512x512 pixels.
  • Speech output operates at a frequency of 24kHz.
  • The model supports a context window exceeding two million tokens.
  • It achieved a score of 93.45 on OmniDocBench V1.5 during private testing.
  • The company aims to open-source its codebase for public use.
  • No revenue model or customer base is currently stated in the facts.

The founders

Abhinav Anand

Founder
  • He is a 19-year-old Class 12 student from Bihar.
  • His father is a government officer and his mother is a housewife.
  • He comes from a middle-class family with no computer science degree.
  • He previously published projects via Lightning AI including a text-to-video system.

In numbers

5.82Bmodel parameters
Rs 11 lakhpersonal savings spent
Rs 64,000GPU compute cost
$35,000seeking funding
512x512image resolution
24kHzspeech output frequency
>2 million tokenscontext window size
93.45OmniDocBench V1.5 score

What happens next

  • Anand plans to seek $35,000 to complete his development pipeline.
  • He intends to release model weights on Hugging Face.
  • The codebase will be made available on GitHub as open source.

Words to know

multimodal AI
Artificial intelligence that can process and understand multiple types of data like text, images, and audio simultaneously.
parameters
Internal variables in a machine learning model that are adjusted during training to improve accuracy.
tokens
Basic units of text that an AI model processes, often representing words or parts of words.
open-source
Software whose source code is made publicly available for anyone to view, modify, and distribute.

The numbers around it

AI & Deep Tech in May 2026: 16 rounds with a disclosed amount, $1.61 billion in all; the biggest was Neysa Networks (USD 1.2 billion). Every sector this year →

Read it as a card in the app

All startup news on 7 May 2026 →

Is this ArcleIntelligence? Get your “Featured on StopDown” badge →