One serving VM
The beta currently runs on one A100-backed serving path. The live dashboard shows the real queue, GPU load, model list, and throughput from that server.
Our mission is to make serious AI coding access normal for Indian students
BharatCode exists because Indian students should not be priced out of the tools that shape professional software work. The public beta is free for everyone, and students who join with a valid .ac.in or .edu.in email keep free access forever.
The economics matter because the promise matters
Free access only works if the compute model is honest. BharatCode is not claiming that GPUs are abundant. We are building a shared, visible resource that can keep serving students by matching demand with capacity, exposing real utilization, and scaling carefully.
What is live today
Founder-funded beta
The current bill covers the A100 path, web app, auth, domain, monitoring, and platform storage while we prove daily student and developer demand.
No hidden capacity
The point of the dashboard is to show the shared resource plainly. If the cluster is busy, users should see that instead of guessing from vague status copy.
How we keep access free
The path is simple: keep the current server useful, learn from real usage, then add capacity in public milestones. Focused usage helps more students share the same compute pool.
Reach 100 daily users
The first milestone is sustained daily usage on the current A100, with enough utilization to justify the next capacity purchase.
Add cheaper clusters
The next stage is purpose-built Indian infrastructure using carefully selected older GPUs where the unit economics can support free student access.
Publish capacity signals
Queue depth, utilization, tokens, and model IDs stay visible so the community can understand the resource it is sharing.
India-first infrastructure
BharatCode is built toward Indian data residency and local infrastructure. The model running on the cluster can change as better open-weight models appear; the strategic commitment is that the serving path and user data stay under an India-first operating model as we scale.
A shared compute commons
Public beta access is free, but the resource is finite. Off-peak usage, thoughtful prompts, and visible stewardship scores help keep the cluster useful without turning the product into an API-key quota maze.