NVIDIA Nemotron API in India

Nemotron is NVIDIA's large open-weight model family, and the Nemotron 3 Ultra model is available in India through unoblox on the same ₹-billed, OpenAI-compatible endpoint used for every other model in the catalogue — one key, one monthly GST invoice, no international card.

What Nemotron brings to the catalogue

Nemotron sits alongside the other large open-weight models on unoblox — Qwen3, Llama 4, Kimi — as an alternative lineage with its own training and alignment choices from NVIDIA. Open-weight models like Nemotron give teams an alternative to closed frontier models for large-scale reasoning and generation workloads, often at a materially different price point, and because it's served through the same endpoint, comparing it against Qwen3 or Llama 4 on a real workload is a one-line model-id change rather than a new integration.

₹ pricing: Nemotron vs other open-weight models

Nemotron's live ₹ rate is on its model page. For comparison, here's where other large open-weight models in the catalogue sit:

Model	Input (₹ / 1M tokens)	Output (₹ / 1M tokens)
Qwen3 235B-A22B	9.07	55.44
Llama 4 Maverick	20.16	80.64
Kimi K2.7	68.5	342.7

Run the same prompt against a couple of these alongside Nemotron before committing a production workload — open-weight models from different labs can behave quite differently on the same task even at similar parameter counts.

Calling Nemotron from the unoblox endpoint

curl https://api.unoblox.ai/v1/chat/completions \
  -H "Authorization: Bearer ub-gw-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nvidia/nemotron-3-ultra-550b-a55b",
    "messages": [{"role": "user", "content": "Outline a three-step plan to reduce support ticket backlog."}]
  }'

Why "open weight" doesn't mean "hosted in India"

Nemotron being open-weight is a licensing and portability fact, not a hosting fact — on unoblox it's served the same way as the other large catalogue models, and unoblox does not claim India data residency for it. What India-specific value unoblox adds here is the same as for every other model: ₹ pricing, a GST invoice with input tax credit, one endpoint, and no international card, so an Indian team can put a large open-weight model into production without setting up separate infrastructure or a foreign billing relationship.

Frequently asked questions

Is NVIDIA Nemotron available in India through unoblox? Yes, nvidia/nemotron-3-ultra-550b-a55b is callable on https://api.unoblox.ai/v1 right after signup, billed in rupees.

What does Nemotron cost on unoblox? See the live ₹ rate on its model page — it isn't in our short fixed-price list in this note.

Is Nemotron hosted on servers in India? Not specifically — it's served the same way as unoblox's other large catalogue models. The India benefit is ₹ billing, GST invoicing and one endpoint, not data residency.

How does Nemotron compare to Qwen3 or Llama 4? All three are large open-weight models with different training lineages; behaviour varies by task, so it's worth testing a real prompt against more than one before choosing.

Can I call Nemotron and a closed model like GPT-5 with the same key? Yes, the same ub-gw- key authenticates every model in the catalogue.

Is there a GST invoice for Nemotron usage? Yes, Nemotron usage is combined with all other model usage into one monthly GST invoice, input tax credit claimable.

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