Google DeepMind

Gemma 4 31B

Type
language model
Context
262K tokens
Max output
262K
Released
2 April 2026
API string
google/gemma-4-31B-it
Weights
open

Prices by provider

Provider Input, $ per 1M Output, $ per 1M In our data since
crof $0.1 $0.3 31 Jul 2026
NanoGPT google/gemma-4-31b-it $0.1 $0.35 31 Jul 2026
OpenRouter $0.1 $0.34 31 Jul 2026
Weights & Biases $0.1 $0.34 31 Jul 2026
LLM Gateway $0.102 $0.297 31 Jul 2026
lilac $0.11 $0.35 31 Jul 2026
Venice AI $0.12 $0.36 1 Aug 2026
DeepInfra $0.13 $0.38 31 Jul 2026
FastRouter $0.13 $0.38 31 Jul 2026
NEAR AI $0.13 $0.4 31 Jul 2026
OrcaRouter $0.13 $0.38 31 Jul 2026
SiliconFlow $0.13 $0.4 31 Jul 2026
friendli $0.14 $0.4 31 Jul 2026
novita-ai $0.14 $0.4 31 Jul 2026
tensormesh $0.14 $0.56 31 Jul 2026
Abacus.AI $0.14 $0.4 31 Jul 2026
Hugging Face $0.14 $0.4 31 Jul 2026
Kilo Code $0.14 $0.4 31 Jul 2026
Merge Gateway $0.14 $0.4 31 Jul 2026
Vercel $0.14 $0.4 31 Jul 2026
libertai $0.15 $0.4 31 Jul 2026
NanoGPT TEE/gemma-4-31b-it $0.15 $0.46 31 Jul 2026
DigitalOcean Gradient AI $0.18 $0.5 31 Jul 2026
aiand $0.2 $0.5 31 Jul 2026
berget $0.275 $0.55 31 Jul 2026
Together AI pearl-ai/gemma-4-31b-it $0.28 $0.86 31 Jul 2026
NanoGPT Gemma-4-31B-it $0.306 $0.306 31 Jul 2026
SambaNova $0.38 $1.15 31 Jul 2026
Together AI google/gemma-4-31B-it $0.39 $0.97 31 Jul 2026
Pioneer $0.5 $0.5 31 Jul 2026

Only the base tier and only per-token prices are shown. Batch, discounted and cached rates, as well as prices per image or per second of video, do not go into this table: they cannot stand in the same column as a price per million tokens.

A single provider appears several times if it sells this model under different identifiers and at different prices — for example, at different weight precision. The identifier is shown next to the name. Identical offers that arrived from two sources under different spellings are merged into one row.

The date in the last column is the day this price first entered our collection. The price may well be older: before that day we simply were not recording it. It has not changed since — otherwise a new row with a new date would stand in its place.

Measurement results

Task set Result Run conditions Measured by
WeirdML — unusual machine learning tasks 52.26 % https://htihle.github.io/weirdml.html
SimpleQA Verified — factual accuracy 9.56 % Epoch evaluations
CritPt — physics problems 1.43 %