Bihar Boy Builds AI Model: 19-Year-Old Spends ₹11 Lakh to Create Multimodal AI, Gets Praised Online
Bihar Boy Develops AI Model: A 19-year-old boy from Bihar has made headlines by building his own AI model. Anand spent ₹11 lakh from his own pocket to build a 582B multimodal AI. The model can generate images at 512×512 resolution and also deliver speech output at 24kHz. The internet has been buzzing with all kinds of reactions to this feat from a middle-class family boy.
A 19-year-old boy from Bihar pooled together ₹11 lakh through sheer hard work and built a massive multimodal AI. The buzz around it online has been so loud that people are directly comparing it to tech giants like Google and OpenAI. Anand, who built this AI model, claims it can generate images at 512×512 resolution and also deliver speech output at 24kHz. Coming from a middle-class family, Anand pulled off this project without any external funding — only his savings and RunPod grants. Anand believes that just like the West or China, India also deserves its own AI.
How Powerful Is Anand’s AI?
According to reports, this multimodal system scored 93.45 on OmniDocBench V1.5 during private testing. Before this project, Anand had already trained a text-to-video system on his own laptop.
Anand’s father is a government officer and his mother is a homemaker. What stands out is that GPU compute alone cost his family around ₹64,000 — a big amount for a middle-class household. He is currently looking for $35,000 in funding for the entire project.
The Internet Can’t Stop Praising Him
Anand’s claims have stirred up the internet and social media, where some people simply cannot stop praising his AI. At the same time, some users are skeptical about how useful this model will actually prove to be. On one side of the internet, people are saying he has put Bihar on the map by building an AI model all on his own. On the other side, some doubt whether this is the result of AI-generated code or vibe coding.
Key Facts
- Anand, a 19-year-old from Bihar, built a 582B multimodal AI model spending ₹11 lakh from personal savings and RunPod grants.
- The model can generate images at 512×512 resolution and deliver speech output at 24kHz.
- During private testing, it scored 93.45 on OmniDocBench V1.5.
- GPU compute alone cost Anand’s middle-class family approximately ₹64,000.
- He is currently seeking $35,000 in funding to advance the project.
- Anand plans to release model weights on Hugging Face and eventually open-source the full codebase on GitHub.
Anand’s Plan Going Forward
Despite all the praise and criticism, Anand remains firm on his goal. According to reports, he plans to release his model weights on Hugging Face and eventually open-source the entire codebase on GitHub.
Many users have praised his ambition — independently training such a large AI model from a state like Bihar is no small thing. Anand’s goal is to plant India firmly on the global AI map, so the country does not have to depend on outside technology.
FAQ
Q: Who is Anand and what exactly did he build?
A: Anand is a 19-year-old from Bihar who built a huge multimodal AI model called 582B entirely on his own. No company backed him. He used personal savings and RunPod grants to do it. The model handles image generation and speech output — the kind of stuff you’d expect from a funded lab, not a teenager from a middle-class family in Bihar.
Q: How much did it cost him to build this AI?
A: He spent around ₹11 lakh in total. GPU computing alone cost his family about ₹64,000, which is a big deal for a middle-class household. He is now looking for $35,000 in outside funding to keep the project moving forward.
Q: How good is this AI model really?
A: In private testing it scored 93.45 on OmniDocBench V1.5. It can generate 512×512 images and produce 24kHz speech output. He had also trained a text-to-video system on his own laptop before this project. Whether it truly competes with Google or OpenAI in real use is still being debated online.
Q: Why are people online doubting him?
A: Some users think the code might be AI-generated or vibe coded rather than written from scratch. Others just find it hard to believe one person could build something this big without a team or proper funding. The skepticism is real — but so is the praise. A lot of people are genuinely impressed by what he pulled off from Bihar.
Q: What does Anand plan to do next with this model?
A: He wants to put the model weights up on Hugging Face so others can access and study them. After that, he plans to open-source the full codebase on GitHub. His bigger goal is to show that India can build its own AI instead of always relying on technology from the West or China.
