Llama 4 API in India

Meta's Llama 4 family — Scout for efficiency, Maverick for power — is live on unoblox's OpenAI-compatible gateway in India, with predictable ₹-native pricing, a GST invoice, and no international licensing to sort out.

Llama 4 pricing

Model	Input (₹/1M)	Output (₹/1M)	Best for
Scout	₹10.08	₹30.24	Cost-first, high-volume
Maverick	₹20.16	₹80.64	Higher-quality reasoning

Open weights mean there's no licensing fee layered on top — you pay only for inference.

Why Llama 4 now

Meta's latest generation closed a lot of the gap with proprietary models on reasoning, coding, and multilingual tasks. At these rates, Llama becomes a strong default for:

• Cost-sensitive agents — multi-turn workflows that need to stay profitable at scale.
• Batch processing — Scout is well-suited to high-volume summarization, classification, and tagging.
• RAG and search — Maverick's extra reasoning depth helps on complex question-answering pipelines.
• Hindi and Indian languages — noticeably better multilingual coverage than earlier Llama generations.

Quick start: one key for both

Base URL: https://api.unoblox.ai/v1
Scout model: meta/llama-4-scout
Maverick model: meta/llama-4-maverick
API Key: ub-gw-[your-key]

Route to either model from the same key — test on Scout first, move to Maverick if you need the extra quality.

Llama for Indian teams

Open-source, no lock-in — you can always run Llama yourself on any cloud; unoblox just removes the DevOps overhead of the API path.

Competitive pricing — Scout's ₹10.08 per million input tokens is one of the more economical rates on the catalog.

GST-compliant invoicing — one monthly rupee bill, input-tax credit claimable.

Frequently asked questions

Can I fine-tune Llama through unoblox? Not through the API today. For custom LoRA adapters, reach out to support to discuss options.

Does Scout run locally too? Yes — Llama's weights are open, so you can download and self-host Scout if you outgrow the managed API.

Which should I choose, Scout or Maverick? Start with Scout — it costs about half of Maverick's input price. Move up to Maverick when answer quality matters more than shaving cost.

Is Llama good at coding? Maverick handles code generation and explanation well; Scout is fine for simpler tasks. Neither is the top choice for the hardest coding problems, but both are competitive for the price.

What's the practical context limit? Check the model card on /models for the current maximum — it's generous enough for most document and chat use cases.

Can I use Llama for production chatbots? Yes — both models are responsive enough for real-time chat: Scout for high-volume simple interactions, Maverick when quality matters more than raw throughput.

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