Gemma vs Llama: Open Model ₹ Pricing Guide

Gemma and Llama are the two big open-weight model families on unoblox — Google's Gemma and Meta's Llama — and both are billed in ₹ per million tokens through the same API key. This page lines up what's actually priced today and how to choose between them.

Two open-weight families, different lineups

Google publishes Gemma as a family of openly available models, and unoblox currently prices Gemma 4 31B directly. A second Gemma variant, Gemma 4 26B-A4B, is also in the catalog but not yet in the fixed ₹ price list below — check its live rate before estimating cost. Meta's Llama 4 family is represented by two sizes, Maverick and Scout, both fully priced.

₹ pricing for what's live today

Model	Model id	Input (₹ / 1M tokens)	Output (₹ / 1M tokens)
Gemma 4 31B	google/gemma-4-31b-it	₹13.10	₹38.30
Gemma 4 26B-A4B	google/gemma-4-26b-a4b-it	see live ₹ pricing	see live ₹ pricing
Llama 4 Scout	meta-llama/llama-4-scout-17b-16e-instruct	₹10.08	₹30.24
Llama 4 Maverick	meta-llama/llama-4-maverick-17b-128e-instruct-fp8	₹20.16	₹80.64

Llama 4 Scout is the cheapest of the four on both input and output tokens, with Gemma 4 31B close behind. Llama 4 Maverick, the larger Llama variant, costs roughly double Scout on both sides of the table.

Picking between them for a real workload

• Cost-first workloads — high-volume classification, tagging, extraction — lean toward Llama 4 Scout or Gemma 4 31B, the two cheapest priced options here.
• Heavier reasoning or generation tasks where you're willing to pay more per token point toward Llama 4 Maverick; validate the step up in quality against your own prompts before committing budget.
• Undecided on Gemma's smaller variant? Check its live model page at https://unoblox.ai/models/google/gemma-4-26b-a4b-it for the current rate rather than assuming it matches the 31B price.

Calling either family from the same key

curl https://api.unoblox.ai/v1/chat/completions \
  -H "Authorization: Bearer ub-gw-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{"model":"google/gemma-4-31b-it","messages":[{"role":"user","content":"Extract the dates mentioned in this email."}]}'

Change "model" to any Llama 4 id above and the request shape stays identical — one endpoint, one key, one GST invoice for both families.

Frequently asked questions

Is Gemma or Llama cheaper on unoblox? Among priced models, Llama 4 Scout is marginally cheaper than Gemma 4 31B on both input and output tokens per million.

Are Gemma and Llama hosted in India? Not by default — both are served from their standard hosting infrastructure through unoblox. The India benefit here is ₹ billing, a GST invoice and one endpoint, not data residency, unless a specific small model is explicitly listed as unoblox-hosted.

What does "open-weight" mean for pricing? It means the model weights are publicly released by Google or Meta; unoblox still meters and bills usage in ₹ per token the same way it does for closed models.

Can I switch from Llama Scout to Llama Maverick without code changes? Yes — only the model field changes in your request; the endpoint, key and response format stay the same.

Why doesn't Gemma 4 26B-A4B have a fixed ₹ price here? Its rate isn't part of the fixed list in this comparison; check its live model page for the current number before budgeting.

Do these models support tool calling and streaming? Yes, through unoblox's OpenAI-compatible endpoint, the same way every other model in the catalog does.

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