Radeon PRO W6800 (32GB)
The Radeon PRO W6800 (32GB) provides 32 GB of memory with 512 GB/s bandwidth, supporting local inference via ROCm and Vulkan acceleration.
Memory
32 GB
Usable Memory Ceiling
32 GB
Bandwidth
512 GB/s
Max Inference Class
32B Models (Q4)
TDP
250W
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 PRO W6800 (32GB)
Inference throughput, memory compatibility, and generation speeds for open-weights LLMs on Radeon PRO W6800 (32GB).
Sort by:
Context:
KV Cache:
GPUs:
| Rank | Model | Parameters | Precision | TPS (Approx.) | TTFT | Can Run |
|---|---|---|---|---|---|---|
#48 | Qwen 3.8 27B | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
#83 | K2 Horizon MoVA 36B A4B | 36B (4B active) | INT4 | 96 tok/s | ~1.9s | |
#93 | Qwen3.5-27B | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
#105 | Sarvam-30B | 32B (2.4B active) | INT4 | 120 tok/s | ~1.3s | |
#106 | Qwen3.6 35B A3B | 35B (3B active) | INT4 | 110 tok/s | ~1.5s | |
#110 | Muse Glimmer 30B | 30B | INT4 Q8 | 23 tok/s 12 tok/s | ~10.6s ~12.5s | |
#111 | MiMo V2 Flash | 15B (309B active) | INT4 Q8 | 3 tok/s 1 tok/s | ~104.2s ~104.5s | |
#112 | Phi-4 Reasoning Plus | 14B | INT4 Q8 FP16 | 34 tok/s 23 tok/s 12 tok/s | ~5.0s ~5.0s ~6.0s | |
#117 | Gemma 4 26B A4B | 25.2B (3.8B active) | INT4 Q8 | 102 tok/s 69 tok/s | ~1.7s ~1.9s | |
#125 | Qwen3.5-9B | 9B | INT4 Q8 FP16 | 64 tok/s 40 tok/s 22 tok/s | ~3.2s ~3.3s ~3.3s | |
#126 | Qwen3.5-35B-A3B | 35B (3B active) | INT4 | 110 tok/s | ~1.5s | |
#128 | K2 Horizon 7B | 7B | INT4 Q8 FP16 | 77 tok/s 50 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
#135 | Gemma 4 12B | 11.95B | INT4 Q8 FP16 | 38 tok/s 26 tok/s 14 tok/s | ~4.3s ~4.3s ~4.7s | |
#140 | Qwen3.5-4B | 4B | INT4 Q8 FP16 | 111 tok/s 76 tok/s 45 tok/s | ~1.5s ~1.5s ~1.5s | |
#143 | GPT-OSS 20B | 21B (3.6B active) | INT4 Q8 | 52 tok/s 43 tok/s | ~1.6s ~1.7s | |
#144 | GLM-4V | 9B | INT4 Q8 FP16 | 47 tok/s 33 tok/s 20 tok/s | ~3.3s ~3.3s ~3.3s | |
#150 | NVIDIA Nemotron 3 Nano 30B-A3B | 3.5B (30B active) | INT4 Q8 FP16 | 22 tok/s 13 tok/s 7 tok/s | ~10.2s ~10.3s ~10.3s | |
#153 | Qwen3-30B-A3B | 30B (3B active) | INT4 Q8 | 111 tok/s 75 tok/s | ~1.5s ~1.8s | |
#158 | Granite 4.2 8B | 8B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~2.9s | |
#159 | Gemma 4 E4B | 8B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~2.9s | |
#163 | Mistral Large 3 | 41B (675B active) | INT4 | 1 tok/s | ~267.5s | |
#165 | GLM-4 | 32B | INT4 | 17 tok/s | ~11.3s | |
#166 | Phi-4 | 14B | INT4 Q8 FP16 | 45 tok/s 27 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
#167 | Gemma 3 27B | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
#170 | DeepSeek-R1 32B | 32B | INT4 Q8 | 23 tok/s 12 tok/s | ~11.3s ~13.3s | |
#171 | Qwen3-32B | 32B | INT4 | 22 tok/s | ~11.3s | |
#175 | Magistral Small | 24B | INT4 Q8 | 22 tok/s 14 tok/s | ~8.5s ~9.3s | |
#176 | OLMo 3.1 32B Instruct | 32B | INT4 | 17 tok/s | ~11.3s | |
#177 | Qwen2.5-32B | 32B | INT4 | 22 tok/s | ~11.3s | |
#178 | OLMo 3 7B Think | 7B | INT4 Q8 FP16 | 57 tok/s 41 tok/s 25 tok/s | ~2.5s ~2.6s ~2.6s | |
#179 | DeepSeek-R1 14B | 14B | INT4 Q8 FP16 | 47 tok/s 28 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
#183 | Ministral 3 14B | 14B | INT4 Q8 FP16 | 34 tok/s 23 tok/s 12 tok/s | ~5.0s ~5.0s ~6.0s | |
#186 | Gemma 3 12B | 12B | INT4 Q8 FP16 | 51 tok/s 31 tok/s 17 tok/s | ~4.3s ~4.3s ~4.3s | |
#188 | Mistral-Small-2501 | 24B | INT4 Q8 | 29 tok/s 16 tok/s | ~8.5s ~9.3s | |
#189 | Ministral 3 8B | 8B | INT4 Q8 FP16 | 52 tok/s 37 tok/s 22 tok/s | ~2.9s ~2.9s ~2.9s | |
#190 | Gemma 4 31B | 30.7B | INT4 Q8 | 23 tok/s 12 tok/s | ~10.8s ~12.7s | |
#191 | MiniCPM5-2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
#192 | Qwen3.5-2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
#193 | Gemma 2 27B | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
#194 | Qwen2.5-14B | 14B | INT4 Q8 FP16 | 45 tok/s 27 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
#195 | OLMo 3.1 32B Think | 32B | INT4 | 17 tok/s | ~11.3s | |
#196 | Nemotron 3.5 Lightning | 30B (3B active) | INT4 | 47 tok/s | ~1.5s | |
#200 | Gemma 4 E2B | 5.1B | INT4 Q8 FP16 | 96 tok/s 64 tok/s 36 tok/s | ~1.9s ~1.9s ~1.9s | |
#202 | DeepSeek-R1 8B | 8B | INT4 Q8 FP16 | 72 tok/s 45 tok/s 25 tok/s | ~2.9s ~2.9s ~2.9s | |
#203 | Kimi Linear 48B A3B Instruct | 48B (3B active) | INT4 | 36 tok/s | ~1.9s | |
#204 | Qwen3-4B | 4B | INT4 Q8 FP16 | 111 tok/s 76 tok/s 45 tok/s | ~1.5s ~1.5s ~1.5s | |
#205 | Ministral 3 3B | 3B | INT4 Q8 FP16 | 91 tok/s 73 tok/s 48 tok/s | ~1.1s ~1.1s ~1.2s | |
#206 | OLMo 3 32B Base | 32B | INT4 | 17 tok/s | ~11.3s | |
#207 | Gemma 2 9B | 9B | INT4 Q8 FP16 | 64 tok/s 40 tok/s 22 tok/s | ~3.2s ~3.3s ~3.3s | |
#208 | Qwen3.5-0.8B | 0.8B | INT4 Q8 FP16 | 214 tok/s 182 tok/s 136 tok/s | ~326 ms ~327 ms ~329 ms | |
#209 | Qwen3-14B | 14B | INT4 Q8 FP16 | 45 tok/s 27 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
#210 | Gemma 3 4B | 4B | INT4 Q8 FP16 | 111 tok/s 76 tok/s 45 tok/s | ~1.5s ~1.5s ~1.5s | |
#211 | Qwen3-8B | 8B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~2.9s | |
#213 | DeepSeek-R1 7B | 7B | INT4 Q8 FP16 | 79 tok/s 50 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
#214 | Qwen3-1.7B | 1.7B | INT4 Q8 FP16 | 167 tok/s 129 tok/s 86 tok/s | ~680 ms ~682 ms ~686 ms | |
#215 | Llama 3.1 8B | 8B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~2.9s | |
#216 | Gemma 3 270M | 0.27B | INT4 Q8 FP16 | 239 tok/s 226 tok/s 198 tok/s | ~112 ms ~112 ms ~113 ms | |
#217 | Command R | 35B | INT4 | 16 tok/s | ~12.3s | |
#218 | Qwen3-0.6B | 0.6B | INT4 Q8 FP16 | 229 tok/s 201 tok/s 157 tok/s | ~243 ms ~243 ms ~245 ms | |
#219 | Llama 3 8B | 8B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~2.9s | |
#221 | Phi-4-Mini | 3.8B | INT4 Q8 FP16 | 114 tok/s 79 tok/s 47 tok/s | ~1.4s ~1.4s ~1.4s | |
#222 | Mixtral-8x7B-v0.1 | 46.7B (7B active) | INT4 | 66 tok/s | ~3.2s | |
#223 | Qwen2.5-7B | 7B | INT4 Q8 FP16 | 77 tok/s 50 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
#224 | Ministral-8B-2410 | 8B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~2.9s | |
#225 | DeepSeek-R1 1.5B | 1.5B | INT4 Q8 FP16 | 175 tok/s 138 tok/s 93 tok/s | ~602 ms ~604 ms ~607 ms | |
#226 | Llama 3.2 3B | 3B | INT4 Q8 FP16 | 130 tok/s 92 tok/s 56 tok/s | ~1.1s ~1.1s ~1.2s | |
#227 | Nemotron 3.5 Lightning 30B A3B | 30B (3B active) | INT4 Q8 | 111 tok/s 75 tok/s | ~1.5s ~1.8s | |
#228 | Phi-3-mini | 3.8B | INT4 Q8 FP16 | 114 tok/s 79 tok/s 47 tok/s | ~1.4s ~1.4s ~1.4s | |
#229 | Gemma 3n E2B IT | 6B (2B active) | INT4 Q8 FP16 | 91 tok/s 79 tok/s 58 tok/s | ~852 ms ~854 ms ~858 ms | |
#230 | Mistral-7B-Instruct-v0.1 | 7.3B | INT4 Q8 FP16 | 75 tok/s 48 tok/s 27 tok/s | ~2.6s ~2.7s ~2.7s | |
#231 | Llama 3.2 1B | 1B | INT4 Q8 FP16 | 200 tok/s 166 tok/s 120 tok/s | ~413 ms ~414 ms ~416 ms | |
#232 | Gemma 2 2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
#233 | Phi-3-medium | 14B | INT4 Q8 FP16 | 45 tok/s 27 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
#234 | OLMo 3 7B Instruct | 7B | INT4 Q8 FP16 | 57 tok/s 41 tok/s 25 tok/s | ~2.5s ~2.6s ~2.6s | |
#235 | Yi-34B | 34B | INT4 | 16 tok/s | ~12.0s | |
#236 | Qwen2-7B | 7B | INT4 Q8 FP16 | 77 tok/s 50 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
#237 | Phi-3-small | 7B | INT4 Q8 FP16 | 77 tok/s 50 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
#238 | Gemma 3 1B | 1B | INT4 Q8 FP16 | 200 tok/s 166 tok/s 120 tok/s | ~413 ms ~414 ms ~416 ms | |
#239 | Mistral-7B-Instruct-v0.2 | 7.3B | INT4 Q8 FP16 | 75 tok/s 48 tok/s 27 tok/s | ~2.6s ~2.7s ~2.7s | |
#241 | Gemma 1 7B | 7B | INT4 Q8 FP16 | 57 tok/s 41 tok/s 25 tok/s | ~2.5s ~2.6s ~2.6s | |
#242 | Gemma 1 2B | 2B | INT4 Q8 FP16 | 164 tok/s 122 tok/s 78 tok/s | ~794 ms ~796 ms ~801 ms | |
#243 | ChatGLM3-6B | 6B | INT4 Q8 FP16 | 63 tok/s 46 tok/s 28 tok/s | ~2.2s ~2.2s ~2.2s | |
#244 | ChatGLM2-6B | 6B | INT4 Q8 FP16 | 63 tok/s 46 tok/s 28 tok/s | ~2.2s ~2.2s ~2.2s | |
#245 | ChatGLM-6B | 6B | INT4 Q8 FP16 | 63 tok/s 46 tok/s 28 tok/s | ~2.2s ~2.2s ~2.2s | |
Unranked Models (Sorted by Speed) | ||||||
- | DINOv3 ViT-L/16 | 0.3B | INT4 Q8 FP16 | 255 tok/s 236 tok/s 203 tok/s | ~124 ms ~124 ms ~125 ms | |
- | Qwen2-0.5B | 0.5B | INT4 Q8 FP16 | 237 tok/s 211 tok/s 170 tok/s | ~202 ms ~203 ms ~204 ms | |
- | Qwen2.5-0.5B | 0.5B | INT4 Q8 FP16 | 237 tok/s 211 tok/s 170 tok/s | ~202 ms ~203 ms ~204 ms | |
- | ERNIE-4.5-0.3B | 0.3B | INT4 Q8 FP16 | 234 tok/s 220 tok/s 191 tok/s | ~124 ms ~124 ms ~125 ms | |
- | ERNIE-4.5-0.3B-Base | 0.3B | INT4 Q8 FP16 | 234 tok/s 220 tok/s 191 tok/s | ~124 ms ~124 ms ~125 ms | |
- | Liquid D1-Omni 600M | 0.6B | INT4 Q8 FP16 | 229 tok/s 201 tok/s 157 tok/s | ~243 ms ~243 ms ~245 ms | |
- | PixArt-α XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 228 tok/s 200 tok/s 156 tok/s | ~246 ms ~247 ms ~248 ms | |
- | PixArt-Σ XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 228 tok/s 200 tok/s 156 tok/s | ~246 ms ~247 ms ~248 ms | |
- | Nemotron 3.5 Parse | 0.63B | INT4 Q8 FP16 | 226 tok/s 197 tok/s 154 tok/s | ~256 ms ~257 ms ~258 ms | |
- | SD 2.1 Turbo | 0.86B | INT4 Q8 FP16 | 210 tok/s 177 tok/s 131 tok/s | ~352 ms ~352 ms ~354 ms | |
- | Stable Diffusion 1.4 | 0.86B | INT4 Q8 FP16 | 210 tok/s 177 tok/s 131 tok/s | ~352 ms ~352 ms ~354 ms | |
- | Stable Diffusion 2.1 | 0.86B | INT4 Q8 FP16 | 210 tok/s 177 tok/s 131 tok/s | ~352 ms ~352 ms ~354 ms | |
- | Falcon-1B | 1B | INT4 Q8 FP16 | 208 tok/s 171 tok/s 123 tok/s | ~413 ms ~414 ms ~416 ms | |
- | Falcon3-1B | 1B | INT4 Q8 FP16 | 200 tok/s 166 tok/s 120 tok/s | ~413 ms ~414 ms ~416 ms | |
- | MiniCPM-V 4.7 1B | 1B | INT4 Q8 FP16 | 200 tok/s 166 tok/s 120 tok/s | ~413 ms ~414 ms ~416 ms | |
- | Kandinsky 2.2 Decoder | 1.22B | INT4 Q8 FP16 | 188 tok/s 152 tok/s 107 tok/s | ~497 ms ~498 ms ~501 ms | |
- | Wan 2.1 T2V 1.3B | 1.3B | INT4 Q8 FP16 | 185 tok/s 148 tok/s 103 tok/s | ~525 ms ~527 ms ~529 ms | |
- | Qwen2-1.5B | 1.5B | INT4 Q8 FP16 | 175 tok/s 138 tok/s 93 tok/s | ~602 ms ~604 ms ~607 ms | |
- | Qwen2.5-1.5B | 1.5B | INT4 Q8 FP16 | 175 tok/s 138 tok/s 93 tok/s | ~602 ms ~604 ms ~607 ms | |
- | SVD Image-to-Video XT | 1.5B | INT4 Q8 FP16 | 175 tok/s 138 tok/s 93 tok/s | ~602 ms ~604 ms ~607 ms | |
- | SVD Image-to-Video XT 1.1 | 1.5B | INT4 Q8 FP16 | 175 tok/s 138 tok/s 93 tok/s | ~602 ms ~604 ms ~607 ms | |
- | LFM3 1.7B | 1.7B | INT4 Q8 FP16 | 167 tok/s 129 tok/s 86 tok/s | ~680 ms ~682 ms ~686 ms | |
- | LFM3 1.7B Base | 1.7B | INT4 Q8 FP16 | 167 tok/s 129 tok/s 86 tok/s | ~680 ms ~682 ms ~686 ms | |
- | Spark X2.5 1.7B | 1.7B | INT4 Q8 FP16 | 167 tok/s 129 tok/s 86 tok/s | ~680 ms ~682 ms ~686 ms | |
- | Hy-MT2 1.8B | 1.8B | INT4 Q8 FP16 | 163 tok/s 125 tok/s 83 tok/s | ~716 ms ~718 ms ~722 ms | |
- | CogVideoX 2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
- | LFM3 Audio 2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
- | LFM3 Omni 2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
- | LFM3 VL 2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
- | Liquid D1 2B | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
- | LTX-Video 0.9.1 | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
- | Moondream 3.1 | 9B (2B active) | INT4 Q8 FP16 | 156 tok/s 118 tok/s 77 tok/s | ~895 ms ~897 ms ~902 ms | |
- | Stable Diffusion 3 Medium | 2B | INT4 Q8 FP16 | 156 tok/s 118 tok/s 76 tok/s | ~794 ms ~796 ms ~801 ms | |
- | Kandinsky 3.0 | 2.1B | INT4 Q8 FP16 | 153 tok/s 115 tok/s 74 tok/s | ~829 ms ~831 ms ~836 ms | |
- | LTX-2 | 2.2B | INT4 Q8 FP16 | 150 tok/s 112 tok/s 71 tok/s | ~864 ms ~866 ms ~871 ms | |
- | SDXL 1.0 Refiner | 2.31B | INT4 Q8 FP16 | 147 tok/s 109 tok/s 69 tok/s | ~902 ms ~904 ms ~910 ms | |
- | Stable Diffusion 3.5 Medium | 2.5B | INT4 Q8 FP16 | 142 tok/s 104 tok/s 65 tok/s | ~968 ms ~971 ms ~976 ms | |
- | SDXL 1.0 Base | 2.56B | INT4 Q8 FP16 | 140 tok/s 102 tok/s 64 tok/s | ~989 ms ~992 ms ~997 ms | |
- | SDXL Turbo | 2.56B | INT4 Q8 FP16 | 140 tok/s 102 tok/s 64 tok/s | ~989 ms ~992 ms ~997 ms | |
- | Falcon-3B | 3B | INT4 Q8 FP16 | 136 tok/s 95 tok/s 58 tok/s | ~1.1s ~1.1s ~1.2s | |
- | CroissantLLM Base | 1.3B | INT4 Q8 FP16 | 134 tok/s 116 tok/s 86 tok/s | ~527 ms ~528 ms ~531 ms | |
- | Phi-1 | 1.3B | INT4 Q8 FP16 | 134 tok/s 116 tok/s 86 tok/s | ~527 ms ~528 ms ~531 ms | |
- | Phi-1.5 | 1.3B | INT4 Q8 FP16 | 134 tok/s 116 tok/s 86 tok/s | ~527 ms ~528 ms ~531 ms | |
- | DeepSeek-R1 3B | 3B | INT4 Q8 FP16 | 131 tok/s 93 tok/s 57 tok/s | ~1.1s ~1.1s ~1.2s | |
- | Mellum 2.1 12B A2.5B Thinking | 12B (2.5B active) | INT4 Q8 FP16 | 131 tok/s 98 tok/s 63 tok/s | ~1.1s ~1.1s ~1.1s | |
- | Falcon3-3B | 3B | INT4 Q8 FP16 | 130 tok/s 92 tok/s 56 tok/s | ~1.1s ~1.1s ~1.2s | |
- | LFM2-VL 3B | 3B | INT4 Q8 FP16 | 130 tok/s 92 tok/s 56 tok/s | ~1.1s ~1.1s ~1.2s | |
- | Liquid D1 3B | 3B | INT4 Q8 FP16 | 130 tok/s 92 tok/s 56 tok/s | ~1.1s ~1.1s ~1.2s | |
- | Liquid D1-VL 3B | 3B | INT4 Q8 FP16 | 130 tok/s 92 tok/s 56 tok/s | ~1.1s ~1.1s ~1.2s | |
- | Liquid LFM2-VL 3B D1 | 3B | INT4 Q8 FP16 | 130 tok/s 92 tok/s 56 tok/s | ~1.1s ~1.1s ~1.2s | |
- | Qwen2.5-3B | 3B | INT4 Q8 FP16 | 130 tok/s 92 tok/s 56 tok/s | ~1.1s ~1.1s ~1.2s | |
- | LFM2.5 24B A2B | 24B (2.5B active) | INT4 Q8 | 122 tok/s 89 tok/s | ~1.3s ~1.4s | |
- | ERNIE-4.5-21B-A3B | 21B (3B active) | INT4 Q8 | 116 tok/s 86 tok/s | ~1.4s ~1.4s | |
- | ERNIE-4.5-21B-A3B-Base | 21B (3B active) | INT4 Q8 | 116 tok/s 86 tok/s | ~1.4s ~1.4s | |
- | ERNIE-4.5-VL-28B-A3B | 28B (3B active) | INT4 Q8 | 112 tok/s 80 tok/s | ~1.5s ~1.6s | |
- | ERNIE-4.5-VL-28B-A3B-Base | 28B (3B active) | INT4 Q8 | 112 tok/s 80 tok/s | ~1.5s ~1.6s | |
- | FLUX.2 Klein 4B | 4B | INT4 Q8 FP16 | 111 tok/s 76 tok/s 45 tok/s | ~1.5s ~1.5s ~1.5s | |
- | FrogNano 4B | 4B | INT4 Q8 FP16 | 111 tok/s 76 tok/s 45 tok/s | ~1.5s ~1.5s ~1.5s | |
- | Hy-Memory 30B A3B | 30B (3B active) | INT4 Q8 | 111 tok/s 75 tok/s | ~1.5s ~1.8s | |
- | Hy-MT2 30B A3B | 30B (3B active) | INT4 Q8 | 111 tok/s 75 tok/s | ~1.5s ~1.8s | |
- | Mage-VL Cowork | 4B | INT4 Q8 FP16 | 111 tok/s 76 tok/s 45 tok/s | ~1.5s ~1.5s ~1.5s | |
- | Spark X2.5 4B | 4B | INT4 Q8 FP16 | 111 tok/s 76 tok/s 45 tok/s | ~1.5s ~1.5s ~1.5s | |
- | MiniCPM-V 4.7 35B A3B | 35B (3B active) | INT4 | 110 tok/s | ~1.5s | |
- | EmbeddingGemma 2 | 2B | INT4 Q8 FP16 | 107 tok/s 89 tok/s 63 tok/s | ~797 ms ~799 ms ~804 ms | |
- | CogVideoX 1.5 5B | 5B | INT4 Q8 FP16 | 97 tok/s 64 tok/s 37 tok/s | ~1.8s ~1.8s ~1.9s | |
- | CogVideoX 5B | 5B | INT4 Q8 FP16 | 97 tok/s 64 tok/s 37 tok/s | ~1.8s ~1.8s ~1.9s | |
- | Wan 2.2 TI2V 5B | 5B | INT4 Q8 FP16 | 97 tok/s 64 tok/s 37 tok/s | ~1.8s ~1.8s ~1.9s | |
- | Phi-2 | 2.7B | INT4 Q8 FP16 | 96 tok/s 77 tok/s 52 tok/s | ~1.0s ~1.0s ~1.0s | |
- | MaLLaM-3B | 3B | INT4 Q8 FP16 | 91 tok/s 73 tok/s 48 tok/s | ~1.1s ~1.1s ~1.2s | |
- | SmolLM3 3B | 3B | INT4 Q8 FP16 | 91 tok/s 73 tok/s 48 tok/s | ~1.1s ~1.1s ~1.2s | |
- | Falcon-7B | 7B | INT4 Q8 FP16 | 81 tok/s 51 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
- | AuraFlow v0.2 | 6.8B | INT4 Q8 FP16 | 79 tok/s 51 tok/s 28 tok/s | ~2.5s ~2.5s ~2.5s | |
- | AuraFlow v0.3 | 6.8B | INT4 Q8 FP16 | 79 tok/s 51 tok/s 28 tok/s | ~2.5s ~2.5s ~2.5s | |
- | Falcon3-7B | 7B | INT4 Q8 FP16 | 77 tok/s 50 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
- | Hy-MT2 7B | 7B | INT4 Q8 FP16 | 77 tok/s 50 tok/s 28 tok/s | ~2.5s ~2.5s ~2.6s | |
- | Mistral-7B-v0.1 | 7.3B | INT4 Q8 FP16 | 75 tok/s 48 tok/s 27 tok/s | ~2.6s ~2.7s ~2.7s | |
- | Youtu Parsing Omni | 8B | INT4 Q8 FP16 | 72 tok/s 45 tok/s 25 tok/s | ~2.9s ~2.9s ~2.9s | |
- | Stable Diffusion 3.5 Large | 8.1B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~3.0s | |
- | Stable Diffusion 3.5 Large Turbo | 8.1B | INT4 Q8 FP16 | 70 tok/s 44 tok/s 24 tok/s | ~2.9s ~2.9s ~3.0s | |
- | FLUX.2 Klein 9B | 9B | INT4 Q8 FP16 | 64 tok/s 40 tok/s 22 tok/s | ~3.2s ~3.3s ~3.3s | |
- | LensVLM 9B | 9B | INT4 Q8 FP16 | 64 tok/s 40 tok/s 22 tok/s | ~3.2s ~3.3s ~3.3s | |
- | MiMo V2.6 Distill Qwen 9B | 9B | INT4 Q8 FP16 | 64 tok/s 40 tok/s 22 tok/s | ~3.2s ~3.3s ~3.3s | |
- | ChatGLM3-6B-32K | 6B | INT4 Q8 FP16 | 63 tok/s 46 tok/s 28 tok/s | ~2.2s ~2.2s ~2.2s | |
- | Yi-6B | 6B | INT4 Q8 FP16 | 63 tok/s 46 tok/s 28 tok/s | ~2.2s ~2.2s ~2.2s | |
- | Kimi-VL-A3B-Instruct | 16B (3B active) | INT4 Q8 | 60 tok/s 52 tok/s | ~1.3s ~1.4s | |
- | Falcon3-10B | 10B | INT4 Q8 FP16 | 59 tok/s 37 tok/s 20 tok/s | ~3.6s ~3.6s ~3.6s | |
- | Mochi 1 Preview | 10B | INT4 Q8 FP16 | 59 tok/s 37 tok/s 20 tok/s | ~3.6s ~3.6s ~3.6s | |
- | Falcon2-11B | 11B | INT4 Q8 FP16 | 57 tok/s 35 tok/s 18 tok/s | ~3.9s ~4.0s ~4.0s | |
- | Hunyuan Lite | 7B | INT4 Q8 FP16 | 57 tok/s 41 tok/s 25 tok/s | ~2.5s ~2.6s ~2.6s | |
- | OLMo 3 7B Base | 7B | INT4 Q8 FP16 | 57 tok/s 41 tok/s 25 tok/s | ~2.5s ~2.6s ~2.6s | |
- | SEA-LION-7B | 7.1B | INT4 Q8 FP16 | 56 tok/s 40 tok/s 24 tok/s | ~2.6s ~2.6s ~2.6s | |
- | SEA-LION-7B-Instruct | 7.1B | INT4 Q8 FP16 | 56 tok/s 40 tok/s 24 tok/s | ~2.6s ~2.6s ~2.6s | |
- | FLUX.1 Canny Dev | 11.9B | INT4 Q8 FP16 | 52 tok/s 31 tok/s 17 tok/s | ~4.3s ~4.3s ~4.3s | |
- | FLUX.1 Depth Dev | 11.9B | INT4 Q8 FP16 | 52 tok/s 31 tok/s 17 tok/s | ~4.3s ~4.3s ~4.3s | |
- | FLUX.1 Dev | 11.9B | INT4 Q8 FP16 | 52 tok/s 31 tok/s 17 tok/s | ~4.3s ~4.3s ~4.3s | |
- | FLUX.1 Fill Dev | 11.9B | INT4 Q8 FP16 | 52 tok/s 31 tok/s 17 tok/s | ~4.3s ~4.3s ~4.3s | |
- | FLUX.1 Schnell | 11.9B | INT4 Q8 FP16 | 52 tok/s 31 tok/s 17 tok/s | ~4.3s ~4.3s ~4.3s | |
- | Sahabat-AI-Llama3-8B-Instruct | 8B | INT4 Q8 FP16 | 52 tok/s 37 tok/s 22 tok/s | ~2.9s ~2.9s ~2.9s | |
- | Typhoon-2-8B | 8B | INT4 Q8 FP16 | 52 tok/s 37 tok/s 22 tok/s | ~2.9s ~2.9s ~2.9s | |
- | HunyuanVideo 1.5 720P T2V | 13B | INT4 Q8 FP16 | 48 tok/s 29 tok/s 15 tok/s | ~4.7s ~4.7s ~5.1s | |
- | GLM-4-9B | 9B | INT4 Q8 FP16 | 47 tok/s 33 tok/s 20 tok/s | ~3.3s ~3.3s ~3.3s | |
- | GLM-4-9B-Chat | 9B | INT4 Q8 FP16 | 47 tok/s 33 tok/s 20 tok/s | ~3.3s ~3.3s ~3.3s | |
- | GLM-4-9B-Chat-1M | 9B | INT4 Q8 FP16 | 47 tok/s 33 tok/s 20 tok/s | ~3.3s ~3.3s ~3.3s | |
- | Sahabat-AI-Gemma2-9B | 9.2B | INT4 Q8 FP16 | 47 tok/s 33 tok/s 19 tok/s | ~3.3s ~3.3s ~3.4s | |
- | Sahabat-AI-Gemma2-9B-Instruct | 9.2B | INT4 Q8 FP16 | 47 tok/s 33 tok/s 19 tok/s | ~3.3s ~3.3s ~3.4s | |
- | Yi-9B | 9B | INT4 Q8 FP16 | 47 tok/s 33 tok/s 20 tok/s | ~3.3s ~3.3s ~3.3s | |
- | Wan 2.1 I2V 14B 720P | 14B | INT4 Q8 FP16 | 45 tok/s 27 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
- | Wan 2.1 T2V 14B | 14B | INT4 Q8 FP16 | 45 tok/s 27 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
- | Wan 2.2 T2V A14B | 14B | INT4 Q8 FP16 | 45 tok/s 27 tok/s 14 tok/s | ~5.0s ~5.0s ~5.5s | |
- | FLUX.2 Dev | 15B | INT4 Q8 FP16 | 43 tok/s 26 tok/s 12 tok/s | ~5.4s ~5.4s ~6.4s | |
- | GreenMind-14B-R1 | 14B | INT4 Q8 FP16 | 34 tok/s 23 tok/s 12 tok/s | ~5.0s ~5.0s ~6.0s | |
- | Arete 27B | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
- | AREX-2 | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
- | PPLX Decider v1 27B | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
- | PPLX Decider v1.1 27B | 27B | INT4 Q8 | 26 tok/s 14 tok/s | ~9.5s ~10.4s | |
- | Agnes 3.0 Flash | 33B | INT4 | 21 tok/s | ~11.6s | |
- | Falcon-40B | 40B | INT4 | 19 tok/s | ~14.0s | |
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 PRO W6800 (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 512 GB/s aggregate bandwidth, batch size 1 inference operates in a memory-bandwidth bound regime. Generates approximately 114 tok/s on an 8B Q4 model and 13 tok/s on a 70B Q4 model.
Frequently Asked Questions
©2025 ApX Machine Learning