GoogleReleased on Apr 2, 2026

Gemma 4 31B: Benchmarks, Pricing & Size

Gemma 4 31B is a language model from Google, released in April 2026, with multimodal input.

Gemma 4 31B is Google DeepMind's flagship dense multimodal model with 31 billion parameters and a 256K context window. Ranks #3 among open models on Arena AI. Built from the same research as Gemini 3, it features Per-Layer Embeddings,

Input
TextImage
Output
Text
Speed
Latency p95851ms
Throughput P9535.6char/s
Cost
$0.16/ 1M · 8:1 in:out
$0.13 in · $0.38 out

Gemma 4 31B pricing

Providers

Gemma 4 31B starts at $0.130 per million input tokens and $0.380 per million output tokens via DeepInfra. See all 4 providers below with their per-token pricing, latency, throughput, and modality support.

ProviderInput $/MOutput $/MMax InputMax OutputLatency p95 sThroughput P95QuantInputOutput
DeepInfra logoDeepInfra
$0.130$0.380262.1K131.1K
0.85
17 c/s
fp8
FriendliAI logoFriendliAI
$0.140$0.400262.1K131.1K
5.92
24 c/s
—
Novita logoNovita
$0.140$0.400262.1K131.1K
0.98
36 c/s
bfloat16
Together logoTogether
$0.390$0.970262.1K131.1K
15.23
—
—
Loading chart...
Loading chart...
Loading chart...

Gemma 4 31B model size

Gemma 4 31B has 30.7 billion parameters. See how it compares to other models in the same parameter range.

Parameters
30.7B
Large (30–80B)
30.7B
1B7B70B405B

Gemma 4 31B API

API access coming soon

Gemma 4 31B will be available through our gateway shortly.

Gemma 4 31B examples

Recent arena outputs from Gemma 4 31B, picked from the highest-ranked matchups.

Gemma 4 31B license

Gemma 4 31B is released under the Apache 2.0 license, which permits commercial use, has 30.7B parameters, has a knowledge cutoff of January 2025.

License
Apache 2.0
Commercial use allowed
Parameters
30.7B
Knowledge cutoff
January 2025

Apache License 2.0 - allows commercial use

FAQ

Common questions about Gemma 4 31B.

When was Gemma 4 31B released?

Gemma 4 31B was released on April 2, 2026 by Google. This is the official Gemma 4 31B release date tracked on LLM Stats.

Is Gemma 4 31B available via API?

Yes, Gemma 4 31B is available via API. See the official documentation for authentication and endpoint details. It is served by 4 providers tracked on LLM Stats.

How big is Gemma 4 31B?

Gemma 4 31B has 30.7 billion parameters. It ships as an open-weight model, so you can download and run it on your own hardware.

Who created Gemma 4 31B?

Gemma 4 31B was created by Google.

What is the license for Gemma 4 31B?

Gemma 4 31B is released under the Apache 2.0 license. This is an open-source / open-weight license that permits self-hosting.

What is the knowledge cutoff date for Gemma 4 31B?

Gemma 4 31B has a knowledge cutoff of January 2025, meaning it was trained on data up to that point and may not know about events after it.

Is Gemma 4 31B multimodal?

Yes, Gemma 4 31B is multimodal and can accept both text and images as input.

Where can I use Gemma 4 31B?

Gemma 4 31B is available through 4 providers including DeepInfra, FriendliAI, Novita, and 1 more.

Where is the Gemma 4 31B paper or technical report?

Gemma 4 31B has a paper or technical report available at https://huggingface.co/blog/gemma4. Use that source for architecture, training, release and evaluation details.