Gemma Pricing in India (₹ per Token)

Google's Gemma line is one of the more budget-friendly open-weight families on unoblox, and like every other model on the catalog it's billed in rupees with a GST tax invoice issued at each top-up — not through an overseas account, and not through a separate billing relationship with Google.

Gemma pricing on unoblox

Model	Input (₹ / 1M tokens)	Output (₹ / 1M tokens)
Gemma 4 31B (google/gemma-4-31b-it)	₹20.24	₹40.48
Gemma 4 26B-A4B (google/gemma-4-26b-a4b-it)	₹0 (free)	₹0 (free)

Gemma 4 31B's rate is mid-range for the catalog — several models, such as DeepSeek V4 Flash and Llama 4 Scout, undercut it. The 26B-A4B variant is currently free (₹0); check /models/google/gemma-4-26b-a4b-it for its live rate before budgeting. Rates are as of 6 October 2026.

A worked example

A request with 3,000 input tokens and 1,000 output tokens on Gemma 4 31B costs: (3,000/1,000,000 × ₹20.24) + (1,000/1,000,000 × ₹40.48) = ₹0.061 + ₹0.040 ≈ ₹0.101. Scaled to 20,000 such requests a month, that's roughly ₹2,024 — a plannable rupee number with no lookup step in between.

Calling Gemma from your existing code

Gemma sits behind the same OpenAI-compatible endpoint as every other model — only the model field changes.

base_url: https://api.unoblox.ai/v1
api_key: ub-gw-xxxxxxxxxxxxxxxxxxxx
model: google/gemma-4-31b-it

Gemma vs other budget open-weight models

• DeepSeek V4 Flash — ₹9.11 / ₹18.21, one of the lowest-priced general models on the catalog.
• Llama 4 Scout — ₹10.12 / ₹30.36, also well below Gemma 4 31B on input price.
• Gemma 4 31B — ₹20.24 / ₹40.48, competitive on both input and output.
• Qwen3 235B-A22B — ₹9.11 / ₹55.67, cheap on input with a larger model behind it.

If your workload is genuinely price-sensitive and quality-tolerant, it's worth benchmarking your own prompts across two or three of these rather than picking on price alone — per-token rate is only half the cost equation; how many tokens a model needs to do the job well is the other half.

Frequently asked questions

Is Gemma 4 31B the cheapest model on unoblox? No. Several models cost less, including DeepSeek V4 Flash and Llama 4 Scout, and Gemma 4 26B-A4B is free. See /models for the full, current list sorted by rate.

Does Gemma run on infrastructure in India? Gemma 4 26B-A4B is self-hosted by unoblox and runs in India. Gemma 4 31B runs on its provider's infrastructure outside India; for that model the India-specific benefit is rupee billing and GST invoicing, not data residency.

Can I use both Gemma variants under one key? Yes — google/gemma-4-31b-it and google/gemma-4-26b-a4b-it are both reachable with the same ub-gw-... key; only the model field changes between calls.

Why is Gemma 4 26B-A4B not priced on this page? Its rate is live on its own model page rather than fixed here, since we only quote numbers we can stand behind at the time of writing. Check /models/google/gemma-4-26b-a4b-it for the current figure.

Is there a free Gemma option for testing? Yes — Gemma 4 26B-A4B is currently free (₹0), and Qwen3.8-27B is also ₹0, so you can prototype before you commit spend to a paid model.

How is Gemma usage billed alongside other models I call? Together, with a GST tax invoice issued at each top-up — there's no separate bill per model or per upstream provider.

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