Open models, served from
our own GPUs in India.
An OpenAI-compatible API over Llama, Qwen, Mistral and more — running on 8× NVIDIA V100 32 GB we own and operate. Prepaid credit in rupees, billed per token, no subscription and no egress surprises.
Trial credit included · no card needed · data stays in India
Plans
Every plan unlocks every model. You are buying prepaid credit, not a tier — the only difference is how much you top up at once.
Starter
Try the platform on real hardware.
- INR 500 of prepaid credit
- Every model on the platform
- One API key
- Email support
Growth
PopularFor a product in active development.
- INR 2,000 of prepaid credit
- Every model on the platform
- IP-restricted keys
- Usage and request logs
- Email support
Scale
For production traffic.
- INR 10,000 of prepaid credit
- Every model on the platform
- IP-restricted keys
- Webhook notifications
- Priority support
Prices exclude GST, which is applied on the invoice. Credit from multiple top-ups shares a single expiry — the furthest one you have bought. Need something larger or a committed rate? Talk to us.
Models and rates
Per million tokens, in rupees. You are charged for what a request actually uses — input and output are priced separately.
| Model | Input / 1M | Output / 1M |
|---|---|---|
llama-3.3-70b |
₹65.00 | ₹78.00 |
qwen3-coder-30b |
₹16.00 | ₹64.00 |
qwen2.5-vl-7b-instruct |
₹21.00 | ₹21.00 |
llama-3.1-8b |
₹11.00 | ₹11.00 |
qwen3-8b |
₹4.50 | ₹15.00 |
bge-m3 |
₹1.10 | — |
all-minilm-l6-v2 |
₹1.50 | — |
Embedding models have no output charge. Rates can change with notice; your panel always shows what you were actually billed, per request.