DeepSeek V4 API in India

DeepSeek's V4 generation is live on unoblox through V4 Flash, the production speed tier, at ₹9.07 per million input tokens and ₹18.14 per million output tokens. Call it through the same OpenAI-compatible endpoint as every other model on the platform, billed monthly in rupees with a GST invoice.

What "V4" means on unoblox today

unoblox currently serves the V4 generation through the Flash endpoint — tuned for throughput without giving up much accuracy. If your workload needs deeper, slower reasoning, DeepSeek V3.2 (₹26.21 / ₹38.30) is the sibling model built for that. Check /models any time a new non-Flash V4 variant goes live — we never invent a price for a model that isn't actually on the catalog yet, and we'd rather send you to the live rate than print a number that could be stale by the time you read it.

If you're migrating from a self-hosted DeepSeek deployment or from a different gateway, the move is usually just the base URL and the model string — the request and response shapes stay OpenAI-compatible, so tool calls, JSON mode, and streaming don't need to be rewritten.

Live ₹ pricing

Model	Input (₹/1M)	Output (₹/1M)	Use case
DeepSeek V4 Flash	₹9.07	₹18.14	Reasoning, math, and coding at speed
DeepSeek V3.2	₹26.21	₹38.30	Deeper, slower reasoning

Pay exactly these rates. No setup fee, no hidden surcharge — just the per-token cost on your monthly invoice.

How to use DeepSeek V4 Flash

1. Get an API key → sign up at https://unoblox.ai/sign-in.
2. Set your base URL → point your OpenAI SDK at https://api.unoblox.ai/v1.
3. Call the model → model: "deepseek-ai/deepseek-v4-flash".

from openai import OpenAI

client = OpenAI(api_key="ub-gw-...", base_url="https://api.unoblox.ai/v1")
response = client.chat.completions.create(
    model="deepseek-ai/deepseek-v4-flash",
    messages=[{"role": "user", "content": "Solve this algebra problem..."}],
)

Benefits on unoblox

• One key, one catalog — switch between DeepSeek, Qwen, GPT, Claude, and Llama with a single string change.
• Billed in rupees — monthly GST invoice, input-tax credit claimable, nothing to reconcile against a foreign statement.
• No vendor lock-in — your integration code stays plain OpenAI-compatible.
• Instant top-ups — add balance through Razorpay and watch it reflect in real time.

Frequently asked questions

Is DeepSeek V4 Flash fast enough for production? Yes — Flash is the tier built for throughput, and it's the default choice for high-volume chat, search, and RAG on unoblox.

Can I use streaming? Yes. Pass stream=true and tokens arrive as they're generated — standard for chat UIs and agents.

Is there a daily request limit? No fixed daily cap. You're bound by your account balance and your key's rate-per-minute tier, which you can raise from the dashboard.

Do I need to change my existing OpenAI code? Only the base URL and the model string. Tool calls, streaming, and JSON mode all work unchanged.

Is my data used to train other customers' models? No. Requests aren't used for model training, and data residency stays in India.

Get started in rupees → https://unoblox.ai/sign-in