ai.canshu.com
中文
VRAM Calculator Advisor & TCO API Pricing Compare Models Leaderboard GPUs

Search models, GPUs, and tools

↑↓ Navigate · Enter Open · ★ Watchlist prioritized ai.canshu.com
Q4_K_M: Tight Fit (Watch Context) 22.47 GB / 23.5 GB Usable

Gemma-4-31B-IT on AMD Radeon RX 7900 XTX 24GB: VRAM Requirement & Speed Estimate

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Gemma-4-31B-IT requires 22.47 GB of VRAM (18.61 GB weights + 2.18 GB KV cache + 1.67 GB overhead), fitting within RX 7900 XTX's 23.5 GB usable memory at ~28.7 tokens/s.

Concurrency (Batch):
Dense 30.7B·60L (d=5376)·32Q/4KV·Ctx 256K·Text+Image+Video
Runnable Models →
GPUs
24 GB VRAM·960 GB/s·Safe Usable 23.5 GB
Tight Fit (Watch Context)22.47 / 23.5 GB~28.7 tok/s
3. Weight Quantization
4. Per-Request Context16K (16,384 tok)
Verified: 2026-11-20·Roofline bandwidth/compute estimate (Batch=1)
Tight Fit (Watch Context)
22.47/ 23.5 GB usable

Physical 24 GB · Load 95.5%

Theoretical Decode Speed
~28.7tokens/s

Actual speed varies ~±25% by runtime

Max Context at Batch=1
28K/ 256K

Enable Q8/Q4 KV quantization to scale batch concurrency

Weights: 18.61 GBKV Cache (1×16K): 2.18 GBOverhead: 1.67 GB
🏃 Smooth & Snappy~17.4s for a 500-token response
Reference Launch CLI:
OLLAMA_CONTEXT_LENGTH=16384 ollama run gemma4:31b

RX 7900 XTX Fits Gemma-4-31B-IT · Cost & TCO

Approx. $899 (~$37/GB VRAM) · 355W TDP

📊 All Quantization Levels for Gemma-4-31B-IT on RX 7900 XTX

Quantization Total VRAM (16K) Verdict Est. Speed Max Context
FP16 / BF16 66.11 GB Out of VRAM (OOM) — —
FP8 / Q8_0 36.75 GB Out of VRAM (OOM) — —
GGUF Q6_K 29.16 GB Partial RAM Offload (Slow) ~9.2 tok/s —
GGUF Q5_K_M 25.75 GB Partial RAM Offload (Slow) ~15.8 tok/s —
Q4_K_M / AWQ 22.47 GB Tight Fit (Watch Context) ~28.7 tok/s 28K
GGUF IQ4_XS 20.12 GB Smooth Fit ~35.3 tok/s 56K
GGUF Q3_K_M 18.79 GB Smooth Fit ~38.2 tok/s 72K
GGUF Q2_K 15.81 GB Smooth Fit ~46.6 tok/s 107K

💡 Recommended Alternative GPUs & Macs for Gemma-4-31B-IT

Frequently Asked Questions

Can AMD Radeon RX 7900 XTX 24GB run Gemma-4-31B-IT locally?

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Gemma-4-31B-IT requires 22.47 GB of VRAM (18.61 GB weights + 2.18 GB KV cache + 1.67 GB overhead), fitting within RX 7900 XTX's 23.5 GB usable memory at ~28.7 tokens/s.

What is the best quantization for Gemma-4-31B-IT on RX 7900 XTX?

The highest-quality quantization that fits cleanly on RX 7900 XTX is Q4_K_M / AWQ (Sweet Spot (Recommended)), using 22.47 GB VRAM at ~28.7 tokens/s with up to 28K context.

How much VRAM does Gemma-4-31B-IT need in full FP16 precision?

In unquantized FP16/BF16 precision (16K context), Gemma-4-31B-IT requires 66.11 GB VRAM (61.4 GB for weights alone).