Can I Run Gemma 4 31B on NVIDIA GeForce RTX 3090?

Autor: Jakub Rusinowski · Ostatnia aktualizacja: 11 września 2026

Yes. Gemma 4 31B needs 18.7 GB at Q4_K_M and the NVIDIA GeForce RTX 3090 has 24 GB, leaving about 5.3 GB of headroom, at roughly 28 tok/s (estimated) with room for up to 16K of context.

See what else this hardware can run →

The numbers

Gemma 4 31B on the NVIDIA GeForce RTX 3090: VRAM by quantization

Assumes a 4K-token context with an f16 KV cache, on an estimated architecture — this model publishes no config we can read. A longer window needs more; a quantized KV cache needs less. “Max context” is the largest window that still fits in 24 GB. Figures are estimates from parameter count, quantization and memory bandwidth — the analyzer lets you tune KV-cache quant and context.

Context costs VRAM too. At Q4_K_M on the NVIDIA GeForce RTX 3090: 8K 21.3 GB · 32K 26.8 GB ✗ · 128K 48.5 GB ✗. Past 32K it no longer fits 24 GB — a q8_0 KV cache buys roughly half of that back, which the calculator will price for you.

What to do next

Nothing to change — run it at Q4_K_M. If you want more capability, check whether the next size up also fits.

No compatible GPU? Gemma 4 31B on 32 GB of system RAM, CPU only: runs comfortably.

Running Gemma 4 31B on this card:

Other Gemma 4 sizes on the NVIDIA GeForce RTX 3090

• Gemma 4 26B-A4B — runs (15.7 GB)
• Gemma 4 12B (Unified) — runs (7.2 GB)
• Gemma 4 E4B — runs (4.5 GB)
• Gemma 4 E2B — runs (2.9 GB)

Gemma 4 31B on nearby hardware

• AMD Radeon RX 7900 XT (20 GB) — runs at a lower quant
• NVIDIA GeForce RTX 5090 (32 GB) — runs

All Gemma 4 sizes on the NVIDIA GeForce RTX 3090 | Gemma 4 31B model page | NVIDIA GeForce RTX 3090 GPU page | Check your hardware