google Open weights Vision Reasoning

Gemma 4 31B pricing & benchmarks

Released Apr 2, 2026 · served by 16 providers · ranked #57 by intelligence in our index

Input / 1M tokens$0.090
Output / 1M tokens$0.340
Context window262K
Cached input
Intelligence index29.4
Coding index43.4
Agentic index14.4
1M in + 300K out$0.19

What Gemma 4 31B is

Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...

Providers and prices for Gemma 4 31B

The same model costs different amounts depending on who serves it. Prices are per 1M tokens, cheapest first. Uptime and throughput are OpenRouter's rolling measurements, not vendor claims.

ProviderInputOutputContext QuantUptime 24hThroughput
DeepInfracheapest $0.090 $0.340 262K fp4 96.6%
CoreWeave $0.100 $0.340 262K bf16 95.0%
OpenInference $0.100 $0.350 262K bf16 98.3%
Venice $0.120 $0.360 256K bf16 99.7%
Chutes $0.120 $0.370 131K fp4 89.9%
DeepInfra $0.130 $0.380 262K fp8 93.7%
SiliconFlow $0.130 $0.400 262K fp8 83.3%
Novita $0.140 $0.400 262K bf16 62.5%
Friendli $0.140 $0.400 262K 99.1%
Morph $0.140 $0.400 175K fp4 98.0%
Crusoe $0.140 $0.400 262K 98.3%
Parasail $0.150 $0.400 262K fp8 97.4%
Phala $0.150 $0.460 262K 96.9%
ModelRun $0.220 $0.550 262K fp4 95.2%
Together $0.390 $0.970 262K 96.3%
SambaNova $0.380 $1.15 131K 98.8%
Cerebras $0.990 $1.49 131K fp16 100.0%

Cheaper models in the same class

Models scoring within 4 points of Gemma 4 31B on the Intelligence Index, but with a lower output price.

FAQ

How much does the Gemma 4 31B API cost?

$0.090 per 1M input tokens and $0.340 per 1M output tokens. A typical workload of 1M input + 300K output tokens costs about $0.19.

Is Gemma 4 31B free?

No. It is a paid model starting at $0.090 per 1M input tokens, though some providers offer trial credits.

Which provider is cheapest for Gemma 4 31B?

DeepInfra at $0.090 input / $0.340 output per 1M tokens (fp4 quantization).

Can I self-host Gemma 4 31B?

Yes — weights are published as google/gemma-4-31B-it on Hugging Face, so you can run it on your own hardware.