FindCostShortlistChangesModels
← All tracked models

Google

Gemma 4 31B API prices by provider

google/gemma-4-31b-it

13 inference providers serve this exact model through OpenRouter. Input and output rates are paired per endpoint; route tags may identify a region or sub-endpoint whose eligibility is not yet verified.

Active endpoints
12 / 15
Lowest at 1M in + 1M out
$0.38
Reka · standard tier
Context up to
262K
Latest price change
None yet
Rates unchanged since first record 2026-09-29
Gemma 4 31B endpoint prices per million tokens
Provider · routeTierInput / 1MOutput / 1MCache readCache writeContextTools1M + 1MStatusPrice since

Reka

reka · unknown

Standard$0.08$0.30$0.05—262KNot declared$0.38ActiveSince

DeepInfra

deepinfra/turbo · fp4

Standard$0.09$0.34$0.05—262KNot declared$0.43ActiveSince

CoreWeave

coreweave/fp4 · fp4

Standard$0.10$0.34$0.10—262KYes$0.44ActiveSince

Venice

venice/fp4 · fp4

Standard$0.12$0.36$0.09—256KYes$0.48ActiveSince

DeepInfra

deepinfra/fp8 · fp8

Standard$0.13$0.38——262KYes$0.51ActiveSince

Crusoe

crusoe/bf16 · bf16

Standard$0.14$0.40$0.14—262KYes$0.54ActiveSince

Friendli

friendli · unknown

Standard$0.14$0.40——262KYes$0.54ActiveSince

Parasail

parasail/fp8 · fp8

Standard$0.15$0.40$0.06—262KYes$0.55ActiveSince

Io Net

io-net · unknown

Standard$0.38$1.15$0.19—262KYes$1.53ActiveSince

SambaNova

sambanova · unknown

Standard$0.38$1.15——262KYes$1.53ActiveSince

ModelRun

modelrun/fp4 · fp4

Standard$0.75$1.00$0.75—262KYes$1.75ActiveSince

SiliconFlow

siliconflow/fp8 · fp8

Standard$0.75$1.00$0.25—262KYes$1.75ActiveSince

Chutes

chutes/fp4 · fp4

Standard$0.12$0.37$0.012—131KYes$0.49UnavailableSince

Novita

novita/bf16 · bf16

Standard$0.14$0.40——262KYes$0.54UnavailableSince

DeepInfra

deepinfra/ultra · fp8

Standard$0.27$0.76——131KYes$1.03UnavailableSince

“—” means no rate is published for that endpoint; it is excluded from estimates, not counted as zero. “Since” is the first record with no rate change observed; “Changed” is the latest observed rate or billing change. Per-request fees: not published for any endpoint here. Status is the upstream operational flag, not a measured uptime.