AMD · RDNA 4 · 2025

AMD Radeon AI PRO R9700 for local AI

32GB for far less than NVIDIA workstation cards and built for stacking, but no faster per card than an RX 9070 XT.

Updated October 14, 2026 · Speeds marked ~ are estimates; seemethodology

VRAM: 32GB
Bandwidth: 640 GB/s
Power: 300W
Runs well: 8 models

R9700

Typically about $1,800

Can the R9700 run LLMs?

The R9700 has 32GB of GDDR6 and 640 GB/s of memory bandwidth, which decides how big a model fits and how fast it generates text. It comfortably runs models up to about 35B parameters: the largest in our catalog is Qwen3.6 35B-A3B at Q5_K_M (28GB with 8K context, ~78 tokens/sec). Our top all-round pick for it is Qwen3.8 27B (Q6_K, ~18 tokens/sec). Llama 3.3 70B Instruct only runs with partial CPU offload, at about 2.4 tokens/sec with CPU offload.

Recommendations

Best models for the R9700

Full list

Best overall · also coding, reasoning, vision
Qwen3.8 27B
Q6_K · 25GB · ~18 tokens/sec
Runs well
Runner-up · also fastest, largest that fits
Qwen3.6 35B-A3B
Q5_K_M · 28GB · ~78 tokens/sec
Runs well

Read: the best LLMs to run on the R9700

Compatibility

What runs on the R9700

Every current model in our catalog, at the best quantization that fits in 32GB with 8K tokens of context. "Partial offload" means part of the model runs from system RAM (we assume 64GB), which is much slower.

Which open-weight models run on the AMD Radeon AI PRO R9700, with the best quantization and estimated speed
Model	Parameters	Best quant	Memory	Speed	Verdict
Gemma 4 E4B	8B	Q8_0	10 GB	~41 tok/s	Runs well
Qwen3.5 9B	9B	Q8_0	12 GB	~36 tok/s	Runs well
Phi-4	14.7B	Q8_0	18 GB	~26 tok/s	Runs well
gpt-oss-20b	20.9B (3.6B active)	MXFP4	15 GB	~75 tok/s	Runs well
Gemma 4 26B A4B	26B (3.8B active)	Q6_K	23 GB	~61 tok/s	Runs well
Qwen3.8 27B	27B	Q6_K	25 GB	~18 tok/s	Runs well
Gemma 4 31B	31B	Q5_K_M	26 GB	~18 tok/s	Runs well
Qwen3.6 35B-A3B	35B (3B active)	Q5_K_M	28 GB	~78 tok/s	Runs well
GLM-4.5-Air	106B (12B active)	Q4_K_M	68 GB	~9.2 tok/s	Partial offload
Llama 4 Scout	109B (17B active)	Q4_K_M	70 GB	~6.6 tok/s	Partial offload
gpt-oss-120b	117B (5.1B active)	MXFP4	68 GB	~23 tok/s	Partial offload
Qwen3.5 122B-A10B	122B (10B active)	Q4_K_M	77 GB	~10 tok/s	Partial offload
DeepSeek V4 Flash	284B (13B active)	—	—	—	Won't fitWhat it needs
GLM-5.3-Flash	320B (18B active)	—	—	—	Won't fitWhat it needs
GLM-5.3	744B (40B active)	—	—	—	Won't fitWhat it needs

Performance

R9700 vs. alternatives on Qwen3.8 27B

Generation speed in tokens per second for a single user. Token generation is limited mostly by memory bandwidth, which is why bandwidth predicts these results so well.

Estimated (method)

Measured

R9700 benchmark results

All benchmarks

Model	Quant	Runtime	Generation	Prompt	Source
Llama 2 7B Chat	Q4_0	llama.cpp (Vulkan)	145 tok/s	5610 tok/s	llama.cpp discussion #10879: Performance of llama.cpp with Vulkan · October 2026
Llama 2 7B Chat	Q4_0	llama.cpp (Vulkan)	125 tok/s	5035 tok/s	llama.cpp discussion #10879: Performance of llama.cpp with Vulkan · October 2026
Llama 2 7B Chat	Q4_0	llama.cpp (ROCm (HIP))	115 tok/s	4665 tok/s	llama.cpp discussion #15021: Performance of llama.cpp on AMD ROCm (HIP) · October 2026
Llama 2 7B Chat	Q4_0	llama.cpp (ROCm (HIP))	94 tok/s	4445 tok/s	llama.cpp discussion #15021: Performance of llama.cpp on AMD ROCm (HIP) · October 2026

Our take

Should you buy the R9700 for AI?

The Radeon AI PRO R9700 uses the full Navi 48 chip with 32GB of GDDR6 on a 256-bit bus for 640 GB/s. Its blower cooler and 300W rating are designed for multi-GPU workstations, and four cards give 128GB of VRAM for large MoE models. ROCm officially supports it, and llama.cpp and vLLM both run on it. AMD set a $1,299 retail price in October 2025, but shortage pricing pushed it to about $1,800 by October 2026.

What we like

32GB per card for less than NVIDIA workstation cards
Blower cooler for dense multi-GPU builds
Official ROCm support

Watch out for

640 GB/s is modest for a 32GB card
Street price well above the $1,299 MSRP
Some CUDA-only tools will not run

Best for: Multi-GPU ROCm workstations · Running 32B models at 6 bit on one card

Specifications

R9700 specs that matter for AI

Memory: 32GB GDDR6
Memory bandwidth: 640 GB/s
Memory bus: 256-bit
Architecture: RDNA 4
Power: 300W
Interface: PCIe 5.0 x16
Released: 2025
Launch price: $1,299
Street price (new): ~$1,800 (2026-10)
Software: ROCm, Vulkan

Sources: Wikipedia, Radeon RX 9000 series, Wikipedia, AMD Radeon AI Pro R9000 specifications table

Cross-shop

Alternatives to the R9700

All GPUs

AMD2025

16GB

GDDR6 · 640 GB/s

RX 9070 XT

~$800

Intel2026

32GB

GDDR6 ECC · 608 GB/s

Arc Pro B70

~$1,300

NVIDIA2025

32GB

GDDR7 · 1,792 GB/s

RTX 5090

~$4,300

Frequently asked questions

How much VRAM does the R9700 have?

The AMD Radeon AI PRO R9700 has 32GB of GDDR6 with 640 GB/s of memory bandwidth on a 256-bit bus.

What is the largest LLM the R9700 can run?

Fully in GPU memory, the largest model in our catalog that runs comfortably is Qwen3.6 35B-A3B (35B (3B active)) at Q5_K_M, using about 28GB with 8K context at ~78 tokens/sec.

Can the R9700 run Llama 3.3 70B?

Only with partial CPU offload. At Q4_K_M the model needs about 47GB with 8K context, so part of it spills into system RAM and speed drops to about 2.4 tokens/sec with CPU offload. For usable 70B speeds you want 48GB of VRAM or more (for example two 24GB cards).

What is the best LLM to run on the R9700?

Our top pick is Qwen3.8 27B at Q6_K, which fits in 32GB with room for 8K of context and runs at ~18 tokens/sec. See our full list of the best models for the R9700 for coding, reasoning, and vision picks.

Is the R9700 good for local AI in 2026?

32GB for far less than NVIDIA workstation cards and built for stacking, but no faster per card than an RX 9070 XT.