# Radeon AI PRO R9700 > OnPrem World · https://onpremworld.com/en/hardware/amd-radeon-ai-pro-r9700 · Last updated: 2026-10-14 · CC BY 4.0 32 GB of VRAM for $1,299 with a blower cooler designed for multi-GPU workstations; 32B models fit at 6-bit. ## Specifications | | | |---|---| | Memory | 32 GB GDDR6 (ECC on Linux only) | | Memory bandwidth | 640 GB/s | | Power | 300 W | | Launch | 2025-07 | | Launch price | US$1,299 | | Architecture | RDNA 4 | | Software | ROCm / Vulkan | ## What it can run (32 GB, 8K) Estimates, not benchmarks. See the methodology. | Model | Parameters | Best quant | Total | tok/s (est.) | |---|---|---|---|---| | Qwen3.5-35B-A3B | 35B | Q6_K | 28.9 GiB | 181 | | Qwen3.6-35B-A3B | 35B | Q6_K | 28.9 GiB | 181 | | Qwen3-32B | 32.8B | Q6_K | 28.9 GiB | 17 | | DeepSeek-R1-Distill-Qwen-32B | 32.76B | Q6_K | 28.9 GiB | 17 | | Gemma 4 31B IT | 30.7B | Q6_K | 26.3 GiB | 18 | | Qwen3-30B-A3B (2507) | 30.5B | Q6_K | 25.8 GiB | 165 | | Qwen3-Coder-30B-A3B-Instruct | 30.5B | Q6_K | 25.8 GiB | 165 | | GLM-4.7-Flash (30B-A3B) | 30B | Q6_K | 25.1 GiB | 181 | | Qwen3.5-27B | 27B | Q8_0 | 29.1 GiB | 16 | | Qwen3.6-27B | 27B | Q8_0 | 29.1 GiB | 16 | | Qwen3.8-27B | 27B | Q8_0 | 29.1 GiB | 16 | | Gemma 3 27B IT | 27B | Q8_0 | 29.6 GiB | 16 | | Gemma 4 26B A4B IT | 25.2B | Q8_0 | 27.0 GiB | 111 | | Mistral Small 3.2 24B Instruct (2506) | 24B | Q8_0 | 26.7 GiB | 18 | | Magistral Small 1.2 (2509) | 24B | Q8_0 | 26.7 GiB | 18 | | Devstral Small 2 24B Instruct (2512) | 24B | Q8_0 | 26.7 GiB | 18 | | gpt-oss-20b | 21B | Q8_0 | 22.5 GiB | 117 | | Phi-4 (14B) | 14B | FP16 | 29.4 GiB | 16 | | Gemma 3 12B IT | 12B | FP16 | 24.8 GiB | 19 | | Gemma 4 12B Unified IT | 11.95B | FP16 | 24.2 GiB | 19 | | Qwen3.5-9B | 9B | FP16 | 18.4 GiB | 25 | | Qwen3-8B | 8.2B | FP16 | 17.7 GiB | 27 | | DeepSeek-R1-0528-Qwen3-8B | 8.19B | FP16 | 17.6 GiB | 27 | | Llama 3.1 8B Instruct | 8B | FP16 | 17.1 GiB | 28 | | Gemma 4 E4B IT | 8B | FP16 | 16.4 GiB | 28 | | Llama 3.2 3B Instruct | 3.21B | FP16 | 7.7 GiB | 70 | | Kimi K2.6 | 1000B | Doesn't fit | – | – | | DeepSeek-R1-0528 (671B) | 671B | Doesn't fit | – | – | | DeepSeek-V4-Flash | 284B | Doesn't fit | – | – | | Qwen3.8-Flash-Next | 180B | Doesn't fit | – | – | | Qwen3.5-122B-A10B | 122B | Doesn't fit | – | – | | Mistral Small 4 119B (2603) | 119B | Doesn't fit | – | – | | gpt-oss-120b | 117B | Doesn't fit | – | – | | Llama 4 Scout 17B-16E Instruct | 109B | Doesn't fit | – | – | | GLM-4.5-Air | 106B | Doesn't fit | – | – | | Qwen3-Coder-Next (80B-A3B) | 80B | Doesn't fit | – | – | | Llama 3.3 70B Instruct | 70B | Doesn't fit | – | – | ## Prices Latest price at each shop, as listed on the day we checked. | Region | Shop | Listing | Price | Per GB | Checked | |---|---|---|---|---|---| | US | Newegg | PowerColor Radeon AI Pro R9700 AI PRO R9700 32G-B 32GB 256-bit GDDR6 PCI Express 5.0 x16 Graphics Card | [US$1,699.99](https://www.newegg.com/powercolor-ai-pro-r9700-32g-b-radeon-ai-pro-r9700-32gb-graphics-card/p/N82E16814131885) | US$53/GB | 2026-10-14 | US value: about 92.4 tok/s on Llama 3.1 8B Instruct (Q4_K_M, 8K), or 5.4 tok/s per US$100. All prices: https://onpremworld.com/en/prices.md ## Frequently asked questions ### Which LLMs can the Radeon AI PRO R9700 run? With 32 GB and an 8K context, the largest models that fit include: Qwen3.5-35B-A3B, Qwen3.6-35B-A3B, Qwen3-32B, DeepSeek-R1-Distill-Qwen-32B, Gemma 4 31B IT. See the full table below. ### How fast is the Radeon AI PRO R9700 for local LLMs? Generation speed is mostly set by memory bandwidth (640 GB/s). For Llama 3.1 8B Instruct at FP16, expect roughly 28 tokens/s; bigger models are slower. ### Can the Radeon AI PRO R9700 run 70B models? Llama 3.3 70B Instruct does not fit on the Radeon AI PRO R9700 (32 GB), even at Q4_K_M with an 8K context. ## Sources - https://www.amd.com/en/products/graphics/workstations/radeon-ai-pro/ai-9000-series/amd-radeon-ai-pro-r9700.html - https://www.amd.com/en/blogs/2025/amd-radeon-ai-pro-r9700-to-be-available-in-workstation.html - https://www.amd.com/en/products/graphics/workstations/radeon-ai-pro.html