google/gemma-3-4b-it vs google/gemma-4-31b-it
For 100,000 calls of 2,000 in and 500 out, google/gemma-3-4b-it is cheaper: $15.00 a month at OpenRouter, against $48.00 for google/gemma-4-31b-it at FriendliAI — 3.2× the cost.
Cheapest on each
google/gemma-3-4b-it
$15.00/mo
cheapest, at OpenRouter — $0.05 in, $0.1 out per million
google/gemma-4-31b-it
$48.00/mo
cheapest, at FriendliAI — $0.14 in, $0.4 out per million
Aligned facts
Same rows on both sides, including the ones we cannot fill.
| Attribute | google/gemma-3-4b-it | google/gemma-4-31b-it |
|---|---|---|
| Context window | 131K | 262K |
| Accepts | Aa · IMG | Aa · IMG · VID |
| Sellers | 1 | 4 |
| Cheapest input rate | $0.05 / M | $0.14 / M |
| Cached input rate | $0.02 / M | $0.1 / M |
No quality or benchmark row appears here. We hold no throughput or accuracy measurement for either model, and a cost-per-quality figure built from a lab’s own claim would look authoritative and mean nothing.
Cost for this workload — both models, every seller
100,000 calls a month · 2,000 input and 500 output tokens each. Sellers of both models lead, cheapest combined cost first.
| Seller | google/gemma-3-4b-it /mo | google/gemma-4-31b-it /mo |
|---|---|---|
| OpenRouter | $15.002026-10-27 | $60.002026-10-27 |
| FriendliAI | ○ no answer | $48.002026-10-21 |
| Together AI | ○ no answer | $126.502026-10-23 |
| SambaNova | ○ no answer | $133.502026-10-22 |
Cheapest tier per seller, this workload only. In and output rates, and a cached prefix, are broken out in the cost calculator.