self-hosted/ai
§01·compatibility · /check

Gemma4 31B on RTX 5090

Yes — Gemma4 31B runs on the RTX 5090 (32 GB). Fastest community-measured result: 3395 prefill tokens/s.

runsmultimodalactive50 series32GB VRAM
model
name
Gemma4 31B
slug
gemma4-31b
vertical
multimodal
status
active
open detail ↗
gpu
name
RTX 5090
slug
rtx-5090
vram
32 GB
series
50
open detail ↗
§02·benchmarks
TaskQuantSpeedVRAMWorksConfidenceSourceVerified
llmQ4_K3395prefill tokens/shardware-corner.net· web2026-05-15
llmQ4_K61.1tokens/shardware-corner.net· web2026-05-15
§03·common questions
Can you run Gemma4 31B on RTX 5090?

Yes — Gemma4 31B runs on the RTX 5090 (32 GB). Fastest community-measured result: 3395 prefill tokens/s.

Which quantizations have been tested for Gemma4 31B on RTX 5090?

Q4_K — measured in community benchmarks.

How fast is Gemma4 31B on RTX 5090?

Up to 3395 prefill tokens/s (llm), the fastest community-measured result.

Are there step-by-step instructions for Gemma4 31B on RTX 5090?

Yes — a step-by-step recipe documents Gemma4 31B on the RTX 5090; see the recipes listed below.

§04·related recipes