AMD · Consumer · RDNA 4 (Navi 48, gfx1201)

AMD Radeon RX 9070 XT

Same Navi 48 chip as the 32 GB Radeon AI PRO R9700, with half the memory.

VRAM

16 GB

GDDR6

Memory bandwidth

640 GB/s

FP16 compute

195 TFLOPS

dense tensor

Price

~$599

new, maker or suggested price

What fits at 4-bit

The 4-bit file plus 8,192 tokens of context and 1.5 GB of overhead, against the VRAM of one card. Speeds are measured medians where someone has published a run at that card count; a blank speed means no run, not a slow one. Machine pages carry estimates for the gaps.

Model	Needs	1 card (16 GB)
Qwen3.5 4B4B · Q4_K_M	4.3 GB	fits
Llama 3.1 8B Instruct8B · Q4_K_M	7.1 GB	fits
Qwen3 8B8.2B · Q4_K_M	7.3 GB	fits
Gemma 4 E4B IT8B · Q4_K_M	6.5 GB	fits
Granite 4.2 8B8.8B · Q4_K_M	7.9 GB	fits
Qwen3.5 9B9B · Q4_K_M	7.0 GB	fits
Gemma 4 12B IT11.95B · Q4_K_M	8.2 GB	fits
Ministral 3 14B Reasoning (2512)13.9B · Q4_K_M	10.4 GB	fits
Phi-414.7B · Q4_K_M	11.3 GB	fits
gpt-oss-20b21B, 3.6B active · MXFP4 (native)	12.5 GB	174 tok/smedian of 1 run

Measured speeds

One card

Single-user runs on 1 card
Model	Quant	Backend	Generation	Prompt	Context	Source
gpt-oss-20b	MXFP4 MoE	llama.cpp (build not stated) Vulkan	174 tok/s	2,754 tok/s2,048-token prompt	0	llama.cpp benchmark discussion2025-11

Specifications

Memory: 16 GB GDDR6
FP16 basis: AMD 'Peak Half Precision (FP16 Matrix)' dense; 389 with structured sparsity.
Board power: 304 W
PCIe: Gen 5 x16
NVLink: No
Cooling: Open-air
Software: ROCm, Vulkan
Released: 2025
Availability: sold new
Price basis: AMD SEP at launch; partner cards only.
Maker page: amd.com

Sources

amd.com: 16 GB GDDR6 256-bit, 640 GB/s, 304 W, PCIe 5.0 x16, 195/389 TFLOPS FP16 matrix
amd.com: SEP $599
rocm.docs.amd.com: ROCm 7.14 supported-GPU list: 7900 XTX/W7900 (gfx1100), 9070 XT/R9700 (gfx1201), MI100 (gfx908) supported; MI50 (gfx906) not supported; Radeon RX cards Ubuntu-only