Radeon AI PRO R9700 (32GB)
The Radeon AI PRO R9700 (32GB) provides 32 GB of memory with 643 GB/s bandwidth, supporting local inference via ROCm and Vulkan acceleration.
Memory
32 GB
Usable Memory Ceiling
32 GB
Bandwidth
643.2 GB/s
Max Inference Class
32B Models (Q4)
TDP
300W
VRAM Scaling by Context Length
Estimated memory requirements as context length increases. Curves crossing above the reference line indicate Out of Memory (OOM).
Precision:
Best Models to Run on Radeon AI PRO R9700 (32GB)
Inference throughput, memory compatibility, and generation speeds for open-weights LLMs on Radeon AI PRO R9700 (32GB).
Sort by:
Context:
KV Cache:
GPUs:
| Rank | Model | Parameters | Precision | TPS (Approx.) | TTFT | Can Run |
|---|---|---|---|---|---|---|
#32 | Qwen 3.8 27B | 27B | INT4 Q8 | 32 tok/s 18 tok/s | ~1.6s ~1.7s | |
#74 | K2 Horizon MoVA 36B A4B | 36B (4B active) | INT4 | 122 tok/s | ~313 ms | |
#82 | Qwen3.5-27B | 27B | INT4 Q8 | 32 tok/s 18 tok/s | ~1.6s ~1.7s | |
#88 | Sarvam-30B | 32B (2.4B active) | INT4 | 153 tok/s | ~220 ms | |
#95 | Muse Glimmer 30B | 30B | INT4 Q8 | 29 tok/s 15 tok/s | ~1.7s ~2.1s | |
#96 | Phi-4 Reasoning Plus | 14B | INT4 Q8 FP16 | 42 tok/s 29 tok/s 15 tok/s | ~838 ms ~849 ms ~1.0s | |
#97 | MiMo V2 Flash | 15B (309B active) | INT4 Q8 | 3 tok/s 2 tok/s | ~17.2s ~17.4s | |
#108 | Qwen3.6 35B A3B | 35B (3B active) | INT4 | 139 tok/s | ~256 ms | |
#110 | Gemma 4 12B | 11.95B | INT4 Q8 FP16 | 48 tok/s 33 tok/s 18 tok/s | ~718 ms ~728 ms ~812 ms | |
#112 | Qwen3.5-35B-A3B | 35B (3B active) | INT4 | 139 tok/s | ~256 ms | |
#116 | Gemma 4 26B A4B | 25.2B (3.8B active) | INT4 Q8 | 129 tok/s 88 tok/s | ~287 ms ~315 ms | |
#118 | K2 Horizon 7B | 7B | INT4 Q8 FP16 | 98 tok/s 63 tok/s 35 tok/s | ~425 ms ~431 ms ~443 ms | |
#130 | Nemotron 3.5 Lightning | 30B (3B active) | INT4 | 59 tok/s | ~261 ms | |
#132 | Qwen3.5-4B | 4B | INT4 Q8 FP16 | 141 tok/s 96 tok/s 56 tok/s | ~252 ms ~255 ms ~262 ms | |
#133 | Qwen3.5-9B | 9B | INT4 Q8 FP16 | 81 tok/s 51 tok/s 28 tok/s | ~541 ms ~549 ms ~565 ms | |
#136 | GPT-OSS 20B | 21B (3.6B active) | INT4 Q8 | 66 tok/s 54 tok/s | ~277 ms ~302 ms | |
#137 | Gemma 4 31B | 30.7B | INT4 Q8 | 29 tok/s 15 tok/s | ~1.8s ~2.1s | |
#141 | NVIDIA Nemotron 3 Nano 30B-A3B | 3.5B (30B active) | INT4 Q8 FP16 | 28 tok/s 17 tok/s 9 tok/s | ~1.7s ~1.7s ~1.8s | |
#143 | Qwen3-30B-A3B | 30B (3B active) | INT4 Q8 | 141 tok/s 95 tok/s | ~251 ms ~297 ms | |
#149 | Granite 4.2 8B | 8B | INT4 Q8 FP16 | 89 tok/s 56 tok/s 31 tok/s | ~482 ms ~489 ms ~504 ms | |
#153 | Nemotron 3.5 Lightning 30B A3B | 30B (3B active) | INT4 Q8 | 141 tok/s 95 tok/s | ~251 ms ~297 ms | |
#155 | Gemma 4 E4B | 8B | INT4 Q8 FP16 | 89 tok/s 56 tok/s 31 tok/s | ~482 ms ~489 ms ~504 ms | |
#156 | Mistral Large 3 | 41B (675B active) | INT4 | 1 tok/s | ~44.0s | |
#158 | GLM-4 | 32B | INT4 | 22 tok/s | ~1.9s | |
#159 | Gemma 3 27B | 27B | INT4 Q8 | 32 tok/s 18 tok/s | ~1.6s ~1.7s | |
#161 | Phi-4 | 14B | INT4 Q8 FP16 | 57 tok/s 34 tok/s 17 tok/s | ~832 ms ~843 ms ~942 ms | |
#163 | GLM-4V | 9B | INT4 Q8 FP16 | 60 tok/s 42 tok/s 25 tok/s | ~546 ms ~553 ms ~569 ms | |
#164 | Qwen3-32B | 32B | INT4 | 28 tok/s | ~1.9s | |
#166 | Qwen2.5-32B | 32B | INT4 | 28 tok/s | ~1.9s | |
#169 | DeepSeek-R1 32B | 32B | INT4 Q8 | 29 tok/s 15 tok/s | ~1.9s ~2.2s | |
#170 | Magistral Small | 24B | INT4 Q8 | 27 tok/s 17 tok/s | ~1.4s ~1.6s | |
#173 | Ministral 3 14B | 14B | INT4 Q8 FP16 | 42 tok/s 29 tok/s 15 tok/s | ~838 ms ~849 ms ~1.0s | |
#174 | OLMo 3.1 32B Instruct | 32B | INT4 | 22 tok/s | ~1.9s | |
#176 | DeepSeek-R1 14B | 14B | INT4 Q8 FP16 | 59 tok/s 35 tok/s 18 tok/s | ~831 ms ~843 ms ~941 ms | |
#177 | Gemma 3 12B | 12B | INT4 Q8 FP16 | 65 tok/s 39 tok/s 21 tok/s | ~716 ms ~726 ms ~748 ms | |
#182 | Mistral-Small-2501 | 24B | INT4 Q8 | 36 tok/s 20 tok/s | ~1.4s ~1.5s | |
#183 | Ministral 3 8B | 8B | INT4 Q8 FP16 | 66 tok/s 46 tok/s 28 tok/s | ~486 ms ~493 ms ~507 ms | |
#184 | Qwen3.5-2B | 2B | INT4 Q8 FP16 | 199 tok/s 150 tok/s 97 tok/s | ~137 ms ~139 ms ~142 ms | |
#185 | Qwen2.5-14B | 14B | INT4 Q8 FP16 | 57 tok/s 34 tok/s 17 tok/s | ~832 ms ~843 ms ~942 ms | |
#186 | MiniCPM5-2B | 2B | INT4 Q8 FP16 | 199 tok/s 150 tok/s 97 tok/s | ~137 ms ~139 ms ~142 ms | |
#187 | Gemma 2 27B | 27B | INT4 Q8 | 32 tok/s 18 tok/s | ~1.6s ~1.7s | |
#188 | OLMo 3 7B Think | 7B | INT4 Q8 FP16 | 72 tok/s 51 tok/s 31 tok/s | ~429 ms ~434 ms ~447 ms | |
#189 | OLMo 3.1 32B Think | 32B | INT4 | 22 tok/s | ~1.9s | |
#190 | Gemma 4 E2B | 5.1B | INT4 Q8 FP16 | 122 tok/s 80 tok/s 46 tok/s | ~315 ms ~319 ms ~329 ms | |
#192 | Ministral 3 3B | 3B | INT4 Q8 FP16 | 116 tok/s 92 tok/s 61 tok/s | ~197 ms ~199 ms ~204 ms | |
#194 | DeepSeek-R1 8B | 8B | INT4 Q8 FP16 | 91 tok/s 57 tok/s 31 tok/s | ~482 ms ~489 ms ~504 ms | |
#195 | Gemma 2 9B | 9B | INT4 Q8 FP16 | 81 tok/s 51 tok/s 28 tok/s | ~541 ms ~549 ms ~565 ms | |
#197 | Gemma 3 4B | 4B | INT4 Q8 FP16 | 141 tok/s 96 tok/s 56 tok/s | ~252 ms ~255 ms ~262 ms | |
#198 | Qwen2.5-7B | 7B | INT4 Q8 FP16 | 98 tok/s 63 tok/s 35 tok/s | ~425 ms ~431 ms ~443 ms | |
#199 | DeepSeek-R1 7B | 7B | INT4 Q8 FP16 | 100 tok/s 63 tok/s 35 tok/s | ~425 ms ~430 ms ~443 ms | |
#199 | Qwen3-14B | 14B | INT4 Q8 FP16 | 57 tok/s 34 tok/s 17 tok/s | ~832 ms ~843 ms ~942 ms | |
#201 | Kimi Linear 48B A3B Instruct | 48B (3B active) | INT4 | 46 tok/s | ~331 ms | |
#202 | Qwen3-8B | 8B | INT4 Q8 FP16 | 89 tok/s 56 tok/s 31 tok/s | ~482 ms ~489 ms ~504 ms | |
#203 | Llama 3.1 8B | 8B | INT4 Q8 FP16 | 89 tok/s 56 tok/s 31 tok/s | ~482 ms ~489 ms ~504 ms | |
#204 | Qwen3-4B | 4B | INT4 Q8 FP16 | 141 tok/s 96 tok/s 56 tok/s | ~252 ms ~255 ms ~262 ms | |
#205 | OLMo 3 32B Base | 32B | INT4 | 22 tok/s | ~1.9s | |
#207 | Command R | 35B | INT4 | 20 tok/s | ~2.0s | |
#209 | Llama 3 8B | 8B | INT4 Q8 FP16 | 89 tok/s 56 tok/s 31 tok/s | ~482 ms ~489 ms ~504 ms | |
#210 | Phi-4-Mini | 3.8B | INT4 Q8 FP16 | 145 tok/s 100 tok/s 59 tok/s | ~240 ms ~243 ms ~250 ms | |
#211 | Gemma 3 270M | 0.27B | INT4 Q8 FP16 | 309 tok/s 290 tok/s 254 tok/s | ~25 ms ~25 ms ~25 ms | |
#212 | Qwen3-1.7B | 1.7B | INT4 Q8 FP16 | 214 tok/s 164 tok/s 109 tok/s | ~118 ms ~120 ms ~123 ms | |
#213 | Qwen3-0.6B | 0.6B | INT4 Q8 FP16 | 295 tok/s 258 tok/s 201 tok/s | ~46 ms ~47 ms ~48 ms | |
#214 | Ministral-8B-2410 | 8B | INT4 Q8 FP16 | 89 tok/s 56 tok/s 31 tok/s | ~482 ms ~489 ms ~504 ms | |
#215 | Mixtral-8x7B-v0.1 | 46.7B (7B active) | INT4 | 83 tok/s | ~533 ms | |
#216 | DeepSeek-R1 1.5B | 1.5B | INT4 Q8 FP16 | 225 tok/s 176 tok/s 119 tok/s | ~105 ms ~107 ms ~109 ms | |
#217 | Llama 3.2 3B | 3B | INT4 Q8 FP16 | 165 tok/s 117 tok/s 71 tok/s | ~194 ms ~197 ms ~202 ms | |
#218 | Phi-3-mini | 3.8B | INT4 Q8 FP16 | 145 tok/s 100 tok/s 59 tok/s | ~240 ms ~243 ms ~250 ms | |
#219 | Qwen3.5-0.8B | 0.8B | INT4 Q8 FP16 | 275 tok/s 233 tok/s 174 tok/s | ~60 ms ~61 ms ~62 ms | |
#220 | Gemma 3n E2B IT | 6B (2B active) | INT4 Q8 FP16 | 116 tok/s 100 tok/s 73 tok/s | ~149 ms ~150 ms ~154 ms | |
#221 | Gemma 2 2B | 2B | INT4 Q8 FP16 | 199 tok/s 150 tok/s 97 tok/s | ~137 ms ~139 ms ~142 ms | |
#222 | Phi-3-medium | 14B | INT4 Q8 FP16 | 57 tok/s 34 tok/s 17 tok/s | ~832 ms ~843 ms ~942 ms | |
#223 | Mistral-7B-Instruct-v0.1 | 7.3B | INT4 Q8 FP16 | 95 tok/s 60 tok/s 33 tok/s | ~442 ms ~448 ms ~461 ms | |
#224 | Llama 3.2 1B | 1B | INT4 Q8 FP16 | 257 tok/s 212 tok/s 153 tok/s | ~74 ms ~75 ms ~77 ms | |
#225 | Yi-34B | 34B | INT4 | 21 tok/s | ~2.0s | |
#226 | OLMo 3 7B Instruct | 7B | INT4 Q8 FP16 | 72 tok/s 51 tok/s 31 tok/s | ~429 ms ~434 ms ~447 ms | |
#227 | Qwen2-7B | 7B | INT4 Q8 FP16 | 98 tok/s 63 tok/s 35 tok/s | ~425 ms ~431 ms ~443 ms | |
#228 | Phi-3-small | 7B | INT4 Q8 FP16 | 98 tok/s 63 tok/s 35 tok/s | ~425 ms ~431 ms ~443 ms | |
#229 | Gemma 3 1B | 1B | INT4 Q8 FP16 | 257 tok/s 212 tok/s 153 tok/s | ~74 ms ~75 ms ~77 ms | |
#230 | Mistral-7B-Instruct-v0.2 | 7.3B | INT4 Q8 FP16 | 95 tok/s 60 tok/s 33 tok/s | ~442 ms ~448 ms ~461 ms | |
#232 | Gemma 1 7B | 7B | INT4 Q8 FP16 | 72 tok/s 51 tok/s 31 tok/s | ~429 ms ~434 ms ~447 ms | |
#233 | Gemma 1 2B | 2B | INT4 Q8 FP16 | 210 tok/s 155 tok/s 99 tok/s | ~137 ms ~138 ms ~142 ms | |
#234 | ChatGLM3-6B | 6B | INT4 Q8 FP16 | 79 tok/s 58 tok/s 35 tok/s | ~371 ms ~375 ms ~386 ms | |
#235 | ChatGLM2-6B | 6B | INT4 Q8 FP16 | 79 tok/s 58 tok/s 35 tok/s | ~371 ms ~375 ms ~386 ms | |
#236 | ChatGLM-6B | 6B | INT4 Q8 FP16 | 79 tok/s 58 tok/s 35 tok/s | ~371 ms ~375 ms ~386 ms | |
Unranked Models (Sorted by Speed) | ||||||
- | Qwen2-0.5B | 0.5B | INT4 Q8 FP16 | 306 tok/s 272 tok/s 218 tok/s | ~39 ms ~40 ms ~41 ms | |
- | Qwen2.5-0.5B | 0.5B | INT4 Q8 FP16 | 306 tok/s 272 tok/s 218 tok/s | ~39 ms ~40 ms ~41 ms | |
- | ERNIE-4.5-0.3B | 0.3B | INT4 Q8 FP16 | 302 tok/s 282 tok/s 245 tok/s | ~27 ms ~27 ms ~28 ms | |
- | ERNIE-4.5-0.3B-Base | 0.3B | INT4 Q8 FP16 | 302 tok/s 282 tok/s 245 tok/s | ~27 ms ~27 ms ~28 ms | |
- | Liquid D1-Omni 600M | 0.6B | INT4 Q8 FP16 | 295 tok/s 258 tok/s 201 tok/s | ~46 ms ~47 ms ~48 ms | |
- | Falcon-1B | 1B | INT4 Q8 FP16 | 267 tok/s 219 tok/s 156 tok/s | ~74 ms ~75 ms ~77 ms | |
- | Falcon3-1B | 1B | INT4 Q8 FP16 | 257 tok/s 212 tok/s 153 tok/s | ~74 ms ~75 ms ~77 ms | |
- | Qwen2-1.5B | 1.5B | INT4 Q8 FP16 | 225 tok/s 176 tok/s 119 tok/s | ~105 ms ~107 ms ~109 ms | |
- | Qwen2.5-1.5B | 1.5B | INT4 Q8 FP16 | 225 tok/s 176 tok/s 119 tok/s | ~105 ms ~107 ms ~109 ms | |
- | Spark X2.5 1.7B | 1.7B | INT4 Q8 FP16 | 214 tok/s 164 tok/s 109 tok/s | ~118 ms ~120 ms ~123 ms | |
- | Hy-MT2 1.8B | 1.8B | INT4 Q8 FP16 | 209 tok/s 160 tok/s 105 tok/s | ~124 ms ~125 ms ~129 ms | |
- | Falcon-3B | 3B | INT4 Q8 FP16 | 173 tok/s 121 tok/s 73 tok/s | ~194 ms ~196 ms ~202 ms | |
- | CroissantLLM Base | 1.3B | INT4 Q8 FP16 | 171 tok/s 147 tok/s 109 tok/s | ~94 ms ~95 ms ~98 ms | |
- | Phi-1 | 1.3B | INT4 Q8 FP16 | 171 tok/s 147 tok/s 109 tok/s | ~94 ms ~95 ms ~98 ms | |
- | Phi-1.5 | 1.3B | INT4 Q8 FP16 | 171 tok/s 147 tok/s 109 tok/s | ~94 ms ~95 ms ~98 ms | |
- | DeepSeek-R1 3B | 3B | INT4 Q8 FP16 | 167 tok/s 118 tok/s 72 tok/s | ~194 ms ~197 ms ~202 ms | |
- | Mellum 2.1 12B A2.5B Thinking | 12B (2.5B active) | INT4 Q8 FP16 | 167 tok/s 125 tok/s 80 tok/s | ~190 ms ~192 ms ~196 ms | |
- | Falcon3-3B | 3B | INT4 Q8 FP16 | 165 tok/s 117 tok/s 71 tok/s | ~194 ms ~197 ms ~202 ms | |
- | LFM2-VL 3B | 3B | INT4 Q8 FP16 | 165 tok/s 117 tok/s 71 tok/s | ~194 ms ~197 ms ~202 ms | |
- | Liquid LFM2-VL 3B D1 | 3B | INT4 Q8 FP16 | 165 tok/s 117 tok/s 71 tok/s | ~194 ms ~197 ms ~202 ms | |
- | Qwen2.5-3B | 3B | INT4 Q8 FP16 | 165 tok/s 117 tok/s 71 tok/s | ~194 ms ~197 ms ~202 ms | |
- | ERNIE-4.5-21B-A3B | 21B (3B active) | INT4 Q8 | 147 tok/s 108 tok/s | ~236 ms ~238 ms | |
- | ERNIE-4.5-21B-A3B-Base | 21B (3B active) | INT4 Q8 | 147 tok/s 108 tok/s | ~236 ms ~238 ms | |
- | ERNIE-4.5-VL-28B-A3B | 28B (3B active) | INT4 Q8 | 143 tok/s 101 tok/s | ~247 ms ~271 ms | |
- | ERNIE-4.5-VL-28B-A3B-Base | 28B (3B active) | INT4 Q8 | 143 tok/s 101 tok/s | ~247 ms ~271 ms | |
- | FrogNano 4B | 4B | INT4 Q8 FP16 | 141 tok/s 96 tok/s 56 tok/s | ~252 ms ~255 ms ~262 ms | |
- | Hy-MT2 30B A3B | 30B (3B active) | INT4 Q8 | 141 tok/s 95 tok/s | ~251 ms ~297 ms | |
- | Spark X2.5 4B | 4B | INT4 Q8 FP16 | 141 tok/s 96 tok/s 56 tok/s | ~252 ms ~255 ms ~262 ms | |
- | EmbeddingGemma 2 | 2B | INT4 Q8 FP16 | 136 tok/s 113 tok/s 80 tok/s | ~139 ms ~141 ms ~144 ms | |
- | Phi-2 | 2.7B | INT4 Q8 FP16 | 121 tok/s 97 tok/s 66 tok/s | ~179 ms ~182 ms ~186 ms | |
- | MaLLaM-3B | 3B | INT4 Q8 FP16 | 116 tok/s 92 tok/s 61 tok/s | ~197 ms ~199 ms ~204 ms | |
- | SmolLM3 3B | 3B | INT4 Q8 FP16 | 116 tok/s 92 tok/s 61 tok/s | ~197 ms ~199 ms ~204 ms | |
- | Falcon-7B | 7B | INT4 Q8 FP16 | 102 tok/s 64 tok/s 35 tok/s | ~424 ms ~430 ms ~443 ms | |
- | Falcon3-7B | 7B | INT4 Q8 FP16 | 98 tok/s 63 tok/s 35 tok/s | ~425 ms ~431 ms ~443 ms | |
- | Hy-MT2 7B | 7B | INT4 Q8 FP16 | 98 tok/s 63 tok/s 35 tok/s | ~425 ms ~431 ms ~443 ms | |
- | Mistral-7B-v0.1 | 7.3B | INT4 Q8 FP16 | 95 tok/s 60 tok/s 33 tok/s | ~442 ms ~448 ms ~461 ms | |
- | LensVLM 9B | 9B | INT4 Q8 FP16 | 81 tok/s 51 tok/s 28 tok/s | ~541 ms ~549 ms ~565 ms | |
- | MiMo V2.6 Distill Qwen 9B | 9B | INT4 Q8 FP16 | 81 tok/s 51 tok/s 28 tok/s | ~541 ms ~549 ms ~565 ms | |
- | ChatGLM3-6B-32K | 6B | INT4 Q8 FP16 | 79 tok/s 58 tok/s 35 tok/s | ~371 ms ~375 ms ~386 ms | |
- | Yi-6B | 6B | INT4 Q8 FP16 | 79 tok/s 58 tok/s 35 tok/s | ~371 ms ~375 ms ~386 ms | |
- | Falcon3-10B | 10B | INT4 Q8 FP16 | 75 tok/s 46 tok/s 25 tok/s | ~599 ms ~607 ms ~625 ms | |
- | Kimi-VL-A3B-Instruct | 16B (3B active) | INT4 Q8 | 75 tok/s 66 tok/s | ~233 ms ~235 ms | |
- | Falcon2-11B | 11B | INT4 Q8 FP16 | 72 tok/s 44 tok/s 23 tok/s | ~656 ms ~665 ms ~685 ms | |
- | Hunyuan Lite | 7B | INT4 Q8 FP16 | 72 tok/s 51 tok/s 31 tok/s | ~429 ms ~434 ms ~447 ms | |
- | OLMo 3 7B Base | 7B | INT4 Q8 FP16 | 72 tok/s 51 tok/s 31 tok/s | ~429 ms ~434 ms ~447 ms | |
- | SEA-LION-7B | 7.1B | INT4 Q8 FP16 | 71 tok/s 51 tok/s 31 tok/s | ~434 ms ~440 ms ~453 ms | |
- | SEA-LION-7B-Instruct | 7.1B | INT4 Q8 FP16 | 71 tok/s 51 tok/s 31 tok/s | ~434 ms ~440 ms ~453 ms | |
- | Sahabat-AI-Llama3-8B-Instruct | 8B | INT4 Q8 FP16 | 66 tok/s 46 tok/s 28 tok/s | ~486 ms ~493 ms ~507 ms | |
- | Typhoon-2-8B | 8B | INT4 Q8 FP16 | 66 tok/s 46 tok/s 28 tok/s | ~486 ms ~493 ms ~507 ms | |
- | GLM-4-9B | 9B | INT4 Q8 FP16 | 60 tok/s 42 tok/s 25 tok/s | ~546 ms ~553 ms ~569 ms | |
- | GLM-4-9B-Chat | 9B | INT4 Q8 FP16 | 60 tok/s 42 tok/s 25 tok/s | ~546 ms ~553 ms ~569 ms | |
- | GLM-4-9B-Chat-1M | 9B | INT4 Q8 FP16 | 60 tok/s 42 tok/s 25 tok/s | ~546 ms ~553 ms ~569 ms | |
- | Yi-9B | 9B | INT4 Q8 FP16 | 60 tok/s 42 tok/s 25 tok/s | ~546 ms ~553 ms ~569 ms | |
- | Sahabat-AI-Gemma2-9B | 9.2B | INT4 Q8 FP16 | 59 tok/s 41 tok/s 24 tok/s | ~557 ms ~564 ms ~581 ms | |
- | Sahabat-AI-Gemma2-9B-Instruct | 9.2B | INT4 Q8 FP16 | 59 tok/s 41 tok/s 24 tok/s | ~557 ms ~564 ms ~581 ms | |
- | GreenMind-14B-R1 | 14B | INT4 Q8 FP16 | 42 tok/s 29 tok/s 15 tok/s | ~838 ms ~849 ms ~1.0s | |
- | AREX-2 | 27B | INT4 Q8 | 32 tok/s 18 tok/s | ~1.6s ~1.7s | |
- | PPLX Decider v1 27B | 27B | INT4 Q8 | 32 tok/s 18 tok/s | ~1.6s ~1.7s | |
- | PPLX Decider v1.1 27B | 27B | INT4 Q8 | 32 tok/s 18 tok/s | ~1.6s ~1.7s | |
- | Agnes 3.0 Flash | 33B | INT4 | 27 tok/s | ~1.9s | |
- | Falcon-40B | 40B | INT4 | 23 tok/s | ~2.3s | |
Inference Capacity
Maximum model parameter class runnable by weight precision.
Context:
4-bit
32B Models
8-bit
14B Models
16-bit
14B Models
Fine-Tuning Capacity
Hardware capacity for local training and adapter fine-tuning.
Context Length:
QLoRA
32B Models
LoRA
14B Models
Full Parameter Training
3B Models
Workload Recommendations
Guidance for optimal precision and operational ceilings on Radeon AI PRO R9700 (32GB).
Inference Sweet Spot
Optimized for 14B models at uncompressed or Q8 precision, and 32B models at Q4 with up to 32k context.
Bandwidth & Speed Profile
With 643.2 GB/s aggregate bandwidth, batch size 1 inference operates in a memory-bandwidth bound regime. Generates approximately 143 tok/s on an 8B Q4 model and 17 tok/s on a 70B Q4 model.
Frequently Asked Questions
©2025 ApX Machine Learning