AMD Radeon RX 9070, AI & Machine Learning Benchmarks & Specs

16GB · AI Score 3.4/100 · anchored estimate vs 51 measured cards

3.4 AI Score Includes estimates

We have not run AMD Radeon RX 9070 on our bench. These figures are anchored estimates, interpolated per workload against the 51 GPUs we did measure (confidence: low). On Llama 3.1 8B (Q4_K_M) AMD Radeon RX 9070 should deliver about 82.5 tokens/sec. Llama 3.3 70B does not fit. It needs roughly 42GB and this card has 16GB. For image generation, SDXL should run near 1.21 it/s, while FLUX.1-dev won't fit at BF16 (needs ~26GB). 6 of the 12 workloads won't fit on 16GB at the tested precision, Qwen3 32B, Llama 3.3 70B, FLUX.1-dev, FLUX.1 Kontext and others. We publish those as hard gates rather than quietly dropping to a smaller quant.

AI & Machine Learning benchmark results

Text Generation tok/s 5

Qwen3 4B126.8

Llama 3.1 8B82.5

Qwen2.5-Coder 14B47.2

Workload	Result	Telemetry	Data
Qwen3 4B	126.8 tok/s	estimated	Est.
Llama 3.1 8B	82.5 tok/s	estimated	Est.
Qwen2.5-Coder 14B	47.2 tok/s	estimated	Est.
Qwen3 32B	✕ Won't fit	VRAM-gated at this precision	Est.
Llama 3.3 70B	✕ Won't fit	VRAM-gated at this precision	Est.

Image Generation images/min 3

Workload	Result	Telemetry	Data
Stable Diffusion XL	2.42 images/min	estimated	Est.
Z-Image Turbo	1.5 images/min	estimated	Est.
FLUX.1 dev	✕ Won't fit	VRAM-gated at this precision	Est.

Image Editing images/min 2

Workload	Result	Telemetry	Data
FLUX.1 Kontext dev	✕ Won't fit	VRAM-gated at this precision	Est.
Qwen-Image-Edit	✕ Won't fit	VRAM-gated at this precision	Est.

Video Generation frames/s 2

Workload	Result	Telemetry	Data
LTX-Video (distilled)	1.33 frames/s	estimated	Est.
Wan 2.2 5B (720p)	✕ Won't fit	VRAM-gated at this precision	Est.

How this estimate is derived. This card hasn’t been through our bench yet, so its numbers are anchored estimates, interpolated from the 51 first-party measured cards (Bandwidth Theil-Sen ladder off NVIDIA anchors × vendor factor (LLM ×0.85, diffusion ×0.5); realistic VRAM floors). The VRAM “won’t fit” gates are exact, since they’re pure capacity limits. Confidence: low. Estimates are replaced with measured data as more silicon goes through the bench. Full methodology →

AMD Radeon RX 9070 specifications

Architecture	RDNA 4 (Navi 48)
Stream processors	3,584
VRAM	16GB GDDR6
Memory bus	256-bit
Memory bandwidth	640 GB/s
Boost clock	2,520 MHz
TDP	220 W
Process	4nm TSMC
Interface	PCIe 5.0 x16
Release date	2025-03-06
Launch MSRP	$549

Verdict, capable, but 16GB sets the ceiling

AMD Radeon RX 9070 scores 3.4/100, #58 of 102. It ran 6 of 12; 6 exceeded its 16GB. Figures are anchored estimates, not measurements, we flag that on every row.

Relative performance: where the AMD Radeon RX 9070 lands

100% = this card, AI & Machine Learning headline metric (AI Score). #21 of 61 desktop cards in this vertical.

GPU	%	AI Score
GeForce RTX 5060 Ti	109%	3.7
AMD Radeon RX 6950 XT	103%	3.5
NVIDIA GeForce RTX 3080 Ti	103%	3.5
AMD Radeon RX 9070 XT	100%	3.4
AMD Radeon RX 9070	100%	3.4
AMD Radeon RX 6800 XT	97%	3.3
AMD Radeon RX 6800	97%	3.3
AMD Radeon RX 6900 XT	97%	3.3
AMD Radeon RX 7800 XT	97%	3.3

One tier up: AMD Radeon RX 9070 XT (0%)
One tier down: AMD Radeon RX 6800 XT (-3%)
Best value alternative ($329): GeForce RTX 5060 Ti (+9%)

Same card, other workloads: AMD Radeon RX 9070 Gaming benchmarks

← All AI & Machine Learning GPU rankings

The silicon

Transistors	53,900 million
Fabricated by	TSMC

Silicon figures from Wikipedia (CC BY-SA 4.0). Benchmarks on this page are our own. Compare every chip.

What this card can build

Whole-job timings, composed from our measured per-model results on this card.

Workflow	Time	Energy	Basis
Full codebase review	21.2 min	n/a	estimate, 0 of 1 stage measured

Can't run: 60-second AI short film (needs Qwen3 32B), 60-second AI short film, narrated (needs Qwen3 32B), 10 short social clips (needs Qwen3 32B), 40-product photo shoot (needs FLUX.1 Kontext dev), 6-panel comic page (needs Qwen3 32B), 20 long-form articles (needs Llama 3.3 70B), Character sheet, 12 poses (needs FLUX.1 dev), 100-photo restoration batch (needs FLUX.1 Kontext dev), 24-frame storyboard (needs Qwen3 32B), 100-photo restore and enlarge (needs FLUX.1 Kontext dev).