AMD RX 7900 XTX (24GB)
The AMD RX 7900 XTX (24GB) provides 24 GB of memory with 960 GB/s bandwidth, supporting local inference via ROCm and Vulkan acceleration.
Memory
24 GB
Usable Memory Ceiling
24 GB
Bandwidth
960 GB/s
Max Inference Class
32B Models (Q4)
TDP
355W
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 AMD RX 7900 XTX (24GB)
Inference throughput, memory compatibility, and generation speeds for open-weights LLMs on AMD RX 7900 XTX (24GB).
Sort by:
Context:
KV Cache:
GPUs:
| Rank | Model | Parameters | Precision | TPS (Approx.) | TTFT | Can Run |
|---|---|---|---|---|---|---|
#48 | Qwen 3.8 27B | 27B | INT4 | 45 tok/s | ~2.7s | |
#83 | K2 Horizon MoVA 36B A4B | 36B (4B active) | INT4 | 132 tok/s | ~588 ms | |
#93 | Qwen3.5-27B | 27B | INT4 | 45 tok/s | ~2.7s | |
#105 | Sarvam-30B | 32B (2.4B active) | INT4 | 165 tok/s | ~380 ms | |
#106 | Qwen3.6 35B A3B | 35B (3B active) | INT4 | 146 tok/s | ~479 ms | |
#110 | Muse Glimmer 30B | 30B | INT4 | 41 tok/s | ~3.0s | |
#111 | MiMo V2 Flash | 15B (309B active) | INT4 Q8 | 5 tok/s 2 tok/s | ~29.9s ~32.6s | |
#112 | Phi-4 Reasoning Plus | 14B | INT4 Q8 | 57 tok/s 40 tok/s | ~1.4s ~1.5s | |
#117 | Gemma 4 26B A4B | 25.2B (3.8B active) | INT4 | 146 tok/s | ~496 ms | |
#125 | Qwen3.5-9B | 9B | INT4 Q8 FP16 | 101 tok/s 67 tok/s 37 tok/s | ~937 ms ~942 ms ~1.0s | |
#126 | Qwen3.5-35B-A3B | 35B (3B active) | INT4 | 146 tok/s | ~479 ms | |
#128 | K2 Horizon 7B | 7B | INT4 Q8 FP16 | 117 tok/s 81 tok/s 48 tok/s | ~735 ms ~739 ms ~748 ms | |
#135 | Gemma 4 12B | 11.95B | INT4 Q8 | 64 tok/s 45 tok/s | ~1.2s ~1.2s | |
#140 | Qwen3.5-4B | 4B | INT4 Q8 FP16 | 155 tok/s 116 tok/s 74 tok/s | ~434 ms ~436 ms ~441 ms | |
#143 | GPT-OSS 20B | 21B (3.6B active) | INT4 | 84 tok/s | ~469 ms | |
#144 | GLM-4V | 9B | INT4 Q8 FP16 | 78 tok/s 57 tok/s 33 tok/s | ~940 ms ~945 ms ~1.0s | |
#150 | NVIDIA Nemotron 3 Nano 30B-A3B | 3.5B (30B active) | INT4 Q8 FP16 | 39 tok/s 24 tok/s 13 tok/s | ~2.9s ~3.0s ~3.0s | |
#153 | Qwen3-30B-A3B | 30B (3B active) | INT4 | 155 tok/s | ~433 ms | |
#158 | Granite 4.2 8B | 8B | INT4 Q8 FP16 | 109 tok/s 73 tok/s 43 tok/s | ~835 ms ~840 ms ~850 ms | |
#159 | Gemma 4 E4B | 8B | INT4 Q8 FP16 | 109 tok/s 73 tok/s 43 tok/s | ~835 ms ~840 ms ~850 ms | |
#165 | GLM-4 | 32B | INT4 | 28 tok/s | ~3.8s | |
#166 | Phi-4 | 14B | INT4 Q8 | 75 tok/s 47 tok/s | ~1.4s ~1.4s | |
#167 | Gemma 3 27B | 27B | INT4 | 45 tok/s | ~2.7s | |
#170 | DeepSeek-R1 32B | 32B | INT4 | 40 tok/s | ~3.2s | |
#171 | Qwen3-32B | 32B | INT4 | 39 tok/s | ~3.2s | |
#175 | Magistral Small | 24B | INT4 | 38 tok/s | ~2.4s | |
#176 | OLMo 3.1 32B Instruct | 32B | INT4 | 28 tok/s | ~3.8s | |
#177 | Qwen2.5-32B | 32B | INT4 | 39 tok/s | ~3.2s | |
#178 | OLMo 3 7B Think | 7B | INT4 Q8 FP16 | 91 tok/s 68 tok/s 43 tok/s | ~738 ms ~741 ms ~750 ms | |
#179 | DeepSeek-R1 14B | 14B | INT4 Q8 | 76 tok/s 48 tok/s | ~1.4s ~1.4s | |
#183 | Ministral 3 14B | 14B | INT4 Q8 | 57 tok/s 40 tok/s | ~1.4s ~1.5s | |
#186 | Gemma 3 12B | 12B | INT4 Q8 FP16 | 83 tok/s 54 tok/s 27 tok/s | ~1.2s ~1.2s ~1.5s | |
#188 | Mistral-Small-2501 | 24B | INT4 | 49 tok/s | ~2.4s | |
#189 | Ministral 3 8B | 8B | INT4 Q8 FP16 | 84 tok/s 62 tok/s 37 tok/s | ~838 ms ~842 ms ~924 ms | |
#190 | Gemma 4 31B | 30.7B | INT4 | 40 tok/s | ~3.1s | |
#191 | MiniCPM5-2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
#192 | Qwen3.5-2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
#193 | Gemma 2 27B | 27B | INT4 | 45 tok/s | ~2.7s | |
#194 | Qwen2.5-14B | 14B | INT4 Q8 | 75 tok/s 47 tok/s | ~1.4s ~1.4s | |
#195 | OLMo 3.1 32B Think | 32B | INT4 | 28 tok/s | ~3.8s | |
#196 | Nemotron 3.5 Lightning | 30B (3B active) | INT4 | 73 tok/s | ~477 ms | |
#200 | Gemma 4 E2B | 5.1B | INT4 Q8 FP16 | 139 tok/s 100 tok/s 61 tok/s | ~544 ms ~547 ms ~554 ms | |
#202 | DeepSeek-R1 8B | 8B | INT4 Q8 FP16 | 111 tok/s 74 tok/s 43 tok/s | ~835 ms ~840 ms ~849 ms | |
#204 | Qwen3-4B | 4B | INT4 Q8 FP16 | 155 tok/s 116 tok/s 74 tok/s | ~434 ms ~436 ms ~441 ms | |
#205 | Ministral 3 3B | 3B | INT4 Q8 FP16 | 134 tok/s 111 tok/s 79 tok/s | ~336 ms ~337 ms ~341 ms | |
#206 | OLMo 3 32B Base | 32B | INT4 | 28 tok/s | ~3.8s | |
#207 | Gemma 2 9B | 9B | INT4 Q8 FP16 | 101 tok/s 67 tok/s 37 tok/s | ~937 ms ~942 ms ~1.0s | |
#208 | Qwen3.5-0.8B | 0.8B | INT4 Q8 FP16 | 242 tok/s 219 tok/s 180 tok/s | ~100 ms ~101 ms ~102 ms | |
#209 | Qwen3-14B | 14B | INT4 Q8 | 75 tok/s 47 tok/s | ~1.4s ~1.4s | |
#210 | Gemma 3 4B | 4B | INT4 Q8 FP16 | 155 tok/s 116 tok/s 74 tok/s | ~434 ms ~436 ms ~441 ms | |
#211 | Qwen3-8B | 8B | INT4 Q8 FP16 | 109 tok/s 73 tok/s 43 tok/s | ~835 ms ~840 ms ~850 ms | |
#213 | DeepSeek-R1 7B | 7B | INT4 Q8 FP16 | 120 tok/s 82 tok/s 48 tok/s | ~735 ms ~739 ms ~747 ms | |
#214 | Qwen3-1.7B | 1.7B | INT4 Q8 FP16 | 207 tok/s 173 tok/s 127 tok/s | ~202 ms ~203 ms ~205 ms | |
#215 | Llama 3.1 8B | 8B | INT4 Q8 FP16 | 109 tok/s 73 tok/s 43 tok/s | ~835 ms ~840 ms ~850 ms | |
#216 | Gemma 3 270M | 0.27B | INT4 Q8 FP16 | 259 tok/s 250 tok/s 231 tok/s | ~39 ms ~39 ms ~39 ms | |
#217 | Command R | 35B | INT4 | 26 tok/s | ~4.1s | |
#218 | Qwen3-0.6B | 0.6B | INT4 Q8 FP16 | 252 tok/s 233 tok/s 199 tok/s | ~76 ms ~77 ms ~77 ms | |
#219 | Llama 3 8B | 8B | INT4 Q8 FP16 | 109 tok/s 73 tok/s 43 tok/s | ~835 ms ~840 ms ~850 ms | |
#221 | Phi-4-Mini | 3.8B | INT4 Q8 FP16 | 159 tok/s 119 tok/s 77 tok/s | ~414 ms ~416 ms ~421 ms | |
#223 | Qwen2.5-7B | 7B | INT4 Q8 FP16 | 117 tok/s 81 tok/s 48 tok/s | ~735 ms ~739 ms ~748 ms | |
#224 | Ministral-8B-2410 | 8B | INT4 Q8 FP16 | 109 tok/s 73 tok/s 43 tok/s | ~835 ms ~840 ms ~850 ms | |
#225 | DeepSeek-R1 1.5B | 1.5B | INT4 Q8 FP16 | 214 tok/s 182 tok/s 136 tok/s | ~179 ms ~180 ms ~182 ms | |
#226 | Llama 3.2 3B | 3B | INT4 Q8 FP16 | 174 tok/s 135 tok/s 90 tok/s | ~334 ms ~335 ms ~339 ms | |
#227 | Nemotron 3.5 Lightning 30B A3B | 30B (3B active) | INT4 | 155 tok/s | ~433 ms | |
#228 | Phi-3-mini | 3.8B | INT4 Q8 FP16 | 159 tok/s 119 tok/s 77 tok/s | ~414 ms ~416 ms ~421 ms | |
#229 | Gemma 3n E2B IT | 6B (2B active) | INT4 Q8 FP16 | 134 tok/s 119 tok/s 92 tok/s | ~252 ms ~253 ms ~255 ms | |
#230 | Mistral-7B-Instruct-v0.1 | 7.3B | INT4 Q8 FP16 | 115 tok/s 78 tok/s 46 tok/s | ~765 ms ~769 ms ~778 ms | |
#231 | Llama 3.2 1B | 1B | INT4 Q8 FP16 | 233 tok/s 206 tok/s 165 tok/s | ~125 ms ~126 ms ~127 ms | |
#232 | Gemma 2 2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
#233 | Phi-3-medium | 14B | INT4 Q8 | 75 tok/s 47 tok/s | ~1.4s ~1.4s | |
#234 | OLMo 3 7B Instruct | 7B | INT4 Q8 FP16 | 91 tok/s 68 tok/s 43 tok/s | ~738 ms ~741 ms ~750 ms | |
#235 | Yi-34B | 34B | INT4 | 26 tok/s | ~4.0s | |
#236 | Qwen2-7B | 7B | INT4 Q8 FP16 | 117 tok/s 81 tok/s 48 tok/s | ~735 ms ~739 ms ~748 ms | |
#237 | Phi-3-small | 7B | INT4 Q8 FP16 | 117 tok/s 81 tok/s 48 tok/s | ~735 ms ~739 ms ~748 ms | |
#238 | Gemma 3 1B | 1B | INT4 Q8 FP16 | 233 tok/s 206 tok/s 165 tok/s | ~125 ms ~126 ms ~127 ms | |
#239 | Mistral-7B-Instruct-v0.2 | 7.3B | INT4 Q8 FP16 | 115 tok/s 78 tok/s 46 tok/s | ~765 ms ~769 ms ~778 ms | |
#241 | Gemma 1 7B | 7B | INT4 Q8 FP16 | 91 tok/s 68 tok/s 43 tok/s | ~738 ms ~741 ms ~750 ms | |
#242 | Gemma 1 2B | 2B | INT4 Q8 FP16 | 205 tok/s 167 tok/s 118 tok/s | ~234 ms ~235 ms ~238 ms | |
#243 | ChatGLM3-6B | 6B | INT4 Q8 FP16 | 99 tok/s 75 tok/s 48 tok/s | ~637 ms ~640 ms ~648 ms | |
#244 | ChatGLM2-6B | 6B | INT4 Q8 FP16 | 99 tok/s 75 tok/s 48 tok/s | ~637 ms ~640 ms ~648 ms | |
#245 | ChatGLM-6B | 6B | INT4 Q8 FP16 | 99 tok/s 75 tok/s 48 tok/s | ~637 ms ~640 ms ~648 ms | |
Unranked Models (Sorted by Speed) | ||||||
- | DINOv3 ViT-L/16 | 0.3B | INT4 Q8 FP16 | 268 tok/s 257 tok/s 235 tok/s | ~42 ms ~42 ms ~43 ms | |
- | Qwen2-0.5B | 0.5B | INT4 Q8 FP16 | 258 tok/s 241 tok/s 210 tok/s | ~65 ms ~65 ms ~66 ms | |
- | Qwen2.5-0.5B | 0.5B | INT4 Q8 FP16 | 258 tok/s 241 tok/s 210 tok/s | ~65 ms ~65 ms ~66 ms | |
- | ERNIE-4.5-0.3B | 0.3B | INT4 Q8 FP16 | 256 tok/s 246 tok/s 226 tok/s | ~42 ms ~43 ms ~43 ms | |
- | ERNIE-4.5-0.3B-Base | 0.3B | INT4 Q8 FP16 | 256 tok/s 246 tok/s 226 tok/s | ~42 ms ~43 ms ~43 ms | |
- | Liquid D1-Omni 600M | 0.6B | INT4 Q8 FP16 | 252 tok/s 233 tok/s 199 tok/s | ~76 ms ~77 ms ~77 ms | |
- | PixArt-α XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 252 tok/s 232 tok/s 198 tok/s | ~77 ms ~78 ms ~78 ms | |
- | PixArt-Σ XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 252 tok/s 232 tok/s 198 tok/s | ~77 ms ~78 ms ~78 ms | |
- | Nemotron 3.5 Parse | 0.63B | INT4 Q8 FP16 | 251 tok/s 231 tok/s 196 tok/s | ~80 ms ~80 ms ~81 ms | |
- | SD 2.1 Turbo | 0.86B | INT4 Q8 FP16 | 239 tok/s 215 tok/s 175 tok/s | ~107 ms ~108 ms ~109 ms | |
- | Stable Diffusion 1.4 | 0.86B | INT4 Q8 FP16 | 239 tok/s 215 tok/s 175 tok/s | ~107 ms ~108 ms ~109 ms | |
- | Stable Diffusion 2.1 | 0.86B | INT4 Q8 FP16 | 239 tok/s 215 tok/s 175 tok/s | ~107 ms ~108 ms ~109 ms | |
- | Falcon-1B | 1B | INT4 Q8 FP16 | 238 tok/s 210 tok/s 167 tok/s | ~125 ms ~126 ms ~127 ms | |
- | Falcon3-1B | 1B | INT4 Q8 FP16 | 233 tok/s 206 tok/s 165 tok/s | ~125 ms ~126 ms ~127 ms | |
- | MiniCPM-V 4.7 1B | 1B | INT4 Q8 FP16 | 233 tok/s 206 tok/s 165 tok/s | ~125 ms ~126 ms ~127 ms | |
- | Kandinsky 2.2 Decoder | 1.22B | INT4 Q8 FP16 | 224 tok/s 195 tok/s 151 tok/s | ~149 ms ~150 ms ~151 ms | |
- | Wan 2.1 T2V 1.3B | 1.3B | INT4 Q8 FP16 | 221 tok/s 191 tok/s 146 tok/s | ~157 ms ~158 ms ~159 ms | |
- | Qwen2-1.5B | 1.5B | INT4 Q8 FP16 | 214 tok/s 182 tok/s 136 tok/s | ~179 ms ~180 ms ~182 ms | |
- | Qwen2.5-1.5B | 1.5B | INT4 Q8 FP16 | 214 tok/s 182 tok/s 136 tok/s | ~179 ms ~180 ms ~182 ms | |
- | SVD Image-to-Video XT | 1.5B | INT4 Q8 FP16 | 214 tok/s 182 tok/s 136 tok/s | ~179 ms ~180 ms ~182 ms | |
- | SVD Image-to-Video XT 1.1 | 1.5B | INT4 Q8 FP16 | 214 tok/s 182 tok/s 136 tok/s | ~179 ms ~180 ms ~182 ms | |
- | LFM3 1.7B | 1.7B | INT4 Q8 FP16 | 207 tok/s 173 tok/s 127 tok/s | ~202 ms ~203 ms ~205 ms | |
- | LFM3 1.7B Base | 1.7B | INT4 Q8 FP16 | 207 tok/s 173 tok/s 127 tok/s | ~202 ms ~203 ms ~205 ms | |
- | Spark X2.5 1.7B | 1.7B | INT4 Q8 FP16 | 207 tok/s 173 tok/s 127 tok/s | ~202 ms ~203 ms ~205 ms | |
- | Hy-MT2 1.8B | 1.8B | INT4 Q8 FP16 | 204 tok/s 170 tok/s 124 tok/s | ~212 ms ~213 ms ~215 ms | |
- | CogVideoX 2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
- | LFM3 Audio 2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
- | LFM3 Omni 2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
- | LFM3 VL 2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
- | Liquid D1 2B | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
- | LTX-Video 0.9.1 | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
- | Moondream 3.1 | 9B (2B active) | INT4 Q8 FP16 | 198 tok/s 163 tok/s 110 tok/s | ~263 ms ~264 ms ~289 ms | |
- | Stable Diffusion 3 Medium | 2B | INT4 Q8 FP16 | 198 tok/s 162 tok/s 116 tok/s | ~234 ms ~235 ms ~238 ms | |
- | Kandinsky 3.0 | 2.1B | INT4 Q8 FP16 | 195 tok/s 159 tok/s 113 tok/s | ~244 ms ~245 ms ~248 ms | |
- | LTX-2 | 2.2B | INT4 Q8 FP16 | 193 tok/s 156 tok/s 110 tok/s | ~254 ms ~255 ms ~258 ms | |
- | SDXL 1.0 Refiner | 2.31B | INT4 Q8 FP16 | 190 tok/s 153 tok/s 107 tok/s | ~265 ms ~266 ms ~269 ms | |
- | Stable Diffusion 3.5 Medium | 2.5B | INT4 Q8 FP16 | 185 tok/s 148 tok/s 102 tok/s | ~284 ms ~285 ms ~288 ms | |
- | SDXL 1.0 Base | 2.56B | INT4 Q8 FP16 | 184 tok/s 146 tok/s 100 tok/s | ~290 ms ~291 ms ~295 ms | |
- | SDXL Turbo | 2.56B | INT4 Q8 FP16 | 184 tok/s 146 tok/s 100 tok/s | ~290 ms ~291 ms ~295 ms | |
- | Falcon-3B | 3B | INT4 Q8 FP16 | 180 tok/s 138 tok/s 92 tok/s | ~334 ms ~335 ms ~339 ms | |
- | CroissantLLM Base | 1.3B | INT4 Q8 FP16 | 178 tok/s 160 tok/s 128 tok/s | ~158 ms ~159 ms ~160 ms | |
- | Phi-1 | 1.3B | INT4 Q8 FP16 | 178 tok/s 160 tok/s 128 tok/s | ~158 ms ~159 ms ~160 ms | |
- | Phi-1.5 | 1.3B | INT4 Q8 FP16 | 178 tok/s 160 tok/s 128 tok/s | ~158 ms ~159 ms ~160 ms | |
- | Mellum 2.1 12B A2.5B Thinking | 12B (2.5B active) | INT4 Q8 FP16 | 176 tok/s 142 tok/s 89 tok/s | ~326 ms ~327 ms ~387 ms | |
- | DeepSeek-R1 3B | 3B | INT4 Q8 FP16 | 175 tok/s 136 tok/s 91 tok/s | ~334 ms ~335 ms ~339 ms | |
- | Falcon3-3B | 3B | INT4 Q8 FP16 | 174 tok/s 135 tok/s 90 tok/s | ~334 ms ~335 ms ~339 ms | |
- | LFM2-VL 3B | 3B | INT4 Q8 FP16 | 174 tok/s 135 tok/s 90 tok/s | ~334 ms ~335 ms ~339 ms | |
- | Liquid D1 3B | 3B | INT4 Q8 FP16 | 174 tok/s 135 tok/s 90 tok/s | ~334 ms ~335 ms ~339 ms | |
- | Liquid D1-VL 3B | 3B | INT4 Q8 FP16 | 174 tok/s 135 tok/s 90 tok/s | ~334 ms ~335 ms ~339 ms | |
- | Liquid LFM2-VL 3B D1 | 3B | INT4 Q8 FP16 | 174 tok/s 135 tok/s 90 tok/s | ~334 ms ~335 ms ~339 ms | |
- | Qwen2.5-3B | 3B | INT4 Q8 FP16 | 174 tok/s 135 tok/s 90 tok/s | ~334 ms ~335 ms ~339 ms | |
- | LFM2.5 24B A2B | 24B (2.5B active) | INT4 | 167 tok/s | ~369 ms | |
- | ERNIE-4.5-21B-A3B | 21B (3B active) | INT4 Q8 | 160 tok/s 114 tok/s | ~406 ms ~478 ms | |
- | ERNIE-4.5-21B-A3B-Base | 21B (3B active) | INT4 Q8 | 160 tok/s 114 tok/s | ~406 ms ~478 ms | |
- | ERNIE-4.5-VL-28B-A3B | 28B (3B active) | INT4 | 156 tok/s | ~427 ms | |
- | ERNIE-4.5-VL-28B-A3B-Base | 28B (3B active) | INT4 | 156 tok/s | ~427 ms | |
- | FLUX.2 Klein 4B | 4B | INT4 Q8 FP16 | 155 tok/s 116 tok/s 74 tok/s | ~434 ms ~436 ms ~441 ms | |
- | FrogNano 4B | 4B | INT4 Q8 FP16 | 155 tok/s 116 tok/s 74 tok/s | ~434 ms ~436 ms ~441 ms | |
- | Hy-Memory 30B A3B | 30B (3B active) | INT4 | 155 tok/s | ~433 ms | |
- | Hy-MT2 30B A3B | 30B (3B active) | INT4 | 155 tok/s | ~433 ms | |
- | Mage-VL Cowork | 4B | INT4 Q8 FP16 | 155 tok/s 116 tok/s 74 tok/s | ~434 ms ~436 ms ~441 ms | |
- | Spark X2.5 4B | 4B | INT4 Q8 FP16 | 155 tok/s 116 tok/s 74 tok/s | ~434 ms ~436 ms ~441 ms | |
- | EmbeddingGemma 2 | 2B | INT4 Q8 FP16 | 151 tok/s 132 tok/s 100 tok/s | ~236 ms ~237 ms ~239 ms | |
- | MiniCPM-V 4.7 35B A3B | 35B (3B active) | INT4 | 146 tok/s | ~479 ms | |
- | CogVideoX 1.5 5B | 5B | INT4 Q8 FP16 | 140 tok/s 101 tok/s 62 tok/s | ~534 ms ~537 ms ~543 ms | |
- | CogVideoX 5B | 5B | INT4 Q8 FP16 | 140 tok/s 101 tok/s 62 tok/s | ~534 ms ~537 ms ~543 ms | |
- | Wan 2.2 TI2V 5B | 5B | INT4 Q8 FP16 | 140 tok/s 101 tok/s 62 tok/s | ~534 ms ~537 ms ~543 ms | |
- | Phi-2 | 2.7B | INT4 Q8 FP16 | 139 tok/s 117 tok/s 84 tok/s | ~306 ms ~307 ms ~310 ms | |
- | MaLLaM-3B | 3B | INT4 Q8 FP16 | 134 tok/s 111 tok/s 79 tok/s | ~336 ms ~337 ms ~341 ms | |
- | SmolLM3 3B | 3B | INT4 Q8 FP16 | 134 tok/s 111 tok/s 79 tok/s | ~336 ms ~337 ms ~341 ms | |
- | Falcon-7B | 7B | INT4 Q8 FP16 | 121 tok/s 82 tok/s 48 tok/s | ~735 ms ~739 ms ~747 ms | |
- | AuraFlow v0.2 | 6.8B | INT4 Q8 FP16 | 119 tok/s 82 tok/s 49 tok/s | ~715 ms ~719 ms ~727 ms | |
- | AuraFlow v0.3 | 6.8B | INT4 Q8 FP16 | 119 tok/s 82 tok/s 49 tok/s | ~715 ms ~719 ms ~727 ms | |
- | Falcon3-7B | 7B | INT4 Q8 FP16 | 117 tok/s 81 tok/s 48 tok/s | ~735 ms ~739 ms ~748 ms | |
- | Hy-MT2 7B | 7B | INT4 Q8 FP16 | 117 tok/s 81 tok/s 48 tok/s | ~735 ms ~739 ms ~748 ms | |
- | Mistral-7B-v0.1 | 7.3B | INT4 Q8 FP16 | 115 tok/s 78 tok/s 46 tok/s | ~765 ms ~769 ms ~778 ms | |
- | Youtu Parsing Omni | 8B | INT4 Q8 FP16 | 111 tok/s 74 tok/s 43 tok/s | ~835 ms ~840 ms ~849 ms | |
- | Stable Diffusion 3.5 Large | 8.1B | INT4 Q8 FP16 | 108 tok/s 73 tok/s 42 tok/s | ~845 ms ~850 ms ~860 ms | |
- | Stable Diffusion 3.5 Large Turbo | 8.1B | INT4 Q8 FP16 | 108 tok/s 73 tok/s 42 tok/s | ~845 ms ~850 ms ~860 ms | |
- | FLUX.2 Klein 9B | 9B | INT4 Q8 FP16 | 101 tok/s 67 tok/s 37 tok/s | ~937 ms ~942 ms ~1.0s | |
- | LensVLM 9B | 9B | INT4 Q8 FP16 | 101 tok/s 67 tok/s 37 tok/s | ~937 ms ~942 ms ~1.0s | |
- | MiMo V2.6 Distill Qwen 9B | 9B | INT4 Q8 FP16 | 101 tok/s 67 tok/s 37 tok/s | ~937 ms ~942 ms ~1.0s | |
- | ChatGLM3-6B-32K | 6B | INT4 Q8 FP16 | 99 tok/s 75 tok/s 48 tok/s | ~637 ms ~640 ms ~648 ms | |
- | Yi-6B | 6B | INT4 Q8 FP16 | 99 tok/s 75 tok/s 48 tok/s | ~637 ms ~640 ms ~648 ms | |
- | Falcon3-10B | 10B | INT4 Q8 FP16 | 94 tok/s 62 tok/s 33 tok/s | ~1.0s ~1.0s ~1.1s | |
- | Kimi-VL-A3B-Instruct | 16B (3B active) | INT4 Q8 | 94 tok/s 80 tok/s | ~395 ms ~429 ms | |
- | Mochi 1 Preview | 10B | INT4 Q8 FP16 | 94 tok/s 62 tok/s 33 tok/s | ~1.0s ~1.0s ~1.1s | |
- | Falcon2-11B | 11B | INT4 Q8 FP16 | 91 tok/s 59 tok/s 29 tok/s | ~1.1s ~1.1s ~1.4s | |
- | Hunyuan Lite | 7B | INT4 Q8 FP16 | 91 tok/s 68 tok/s 43 tok/s | ~738 ms ~741 ms ~750 ms | |
- | OLMo 3 7B Base | 7B | INT4 Q8 FP16 | 91 tok/s 68 tok/s 43 tok/s | ~738 ms ~741 ms ~750 ms | |
- | SEA-LION-7B | 7.1B | INT4 Q8 FP16 | 90 tok/s 67 tok/s 42 tok/s | ~747 ms ~751 ms ~760 ms | |
- | SEA-LION-7B-Instruct | 7.1B | INT4 Q8 FP16 | 90 tok/s 67 tok/s 42 tok/s | ~747 ms ~751 ms ~760 ms | |
- | FLUX.1 Canny Dev | 11.9B | INT4 Q8 FP16 | 84 tok/s 54 tok/s 27 tok/s | ~1.2s ~1.2s ~1.5s | |
- | FLUX.1 Depth Dev | 11.9B | INT4 Q8 FP16 | 84 tok/s 54 tok/s 27 tok/s | ~1.2s ~1.2s ~1.5s | |
- | FLUX.1 Dev | 11.9B | INT4 Q8 FP16 | 84 tok/s 54 tok/s 27 tok/s | ~1.2s ~1.2s ~1.5s | |
- | FLUX.1 Fill Dev | 11.9B | INT4 Q8 FP16 | 84 tok/s 54 tok/s 27 tok/s | ~1.2s ~1.2s ~1.5s | |
- | FLUX.1 Schnell | 11.9B | INT4 Q8 FP16 | 84 tok/s 54 tok/s 27 tok/s | ~1.2s ~1.2s ~1.5s | |
- | Sahabat-AI-Llama3-8B-Instruct | 8B | INT4 Q8 FP16 | 84 tok/s 62 tok/s 37 tok/s | ~838 ms ~842 ms ~924 ms | |
- | Typhoon-2-8B | 8B | INT4 Q8 FP16 | 84 tok/s 62 tok/s 37 tok/s | ~838 ms ~842 ms ~924 ms | |
- | HunyuanVideo 1.5 720P T2V | 13B | INT4 Q8 | 79 tok/s 50 tok/s | ~1.3s ~1.3s | |
- | GLM-4-9B | 9B | INT4 Q8 FP16 | 78 tok/s 57 tok/s 33 tok/s | ~940 ms ~945 ms ~1.0s | |
- | GLM-4-9B-Chat | 9B | INT4 Q8 FP16 | 78 tok/s 57 tok/s 33 tok/s | ~940 ms ~945 ms ~1.0s | |
- | GLM-4-9B-Chat-1M | 9B | INT4 Q8 FP16 | 78 tok/s 57 tok/s 33 tok/s | ~940 ms ~945 ms ~1.0s | |
- | Yi-9B | 9B | INT4 Q8 FP16 | 78 tok/s 57 tok/s 33 tok/s | ~940 ms ~945 ms ~1.0s | |
- | Sahabat-AI-Gemma2-9B | 9.2B | INT4 Q8 FP16 | 77 tok/s 56 tok/s 33 tok/s | ~960 ms ~965 ms ~1.1s | |
- | Sahabat-AI-Gemma2-9B-Instruct | 9.2B | INT4 Q8 FP16 | 77 tok/s 56 tok/s 33 tok/s | ~960 ms ~965 ms ~1.1s | |
- | Wan 2.1 I2V 14B 720P | 14B | INT4 Q8 | 75 tok/s 47 tok/s | ~1.4s ~1.4s | |
- | Wan 2.1 T2V 14B | 14B | INT4 Q8 | 75 tok/s 47 tok/s | ~1.4s ~1.4s | |
- | Wan 2.2 T2V A14B | 14B | INT4 Q8 | 75 tok/s 47 tok/s | ~1.4s ~1.4s | |
- | FLUX.2 Dev | 15B | INT4 Q8 | 71 tok/s 45 tok/s | ~1.5s ~1.6s | |
- | GreenMind-14B-R1 | 14B | INT4 Q8 | 57 tok/s 40 tok/s | ~1.4s ~1.5s | |
- | Arete 27B | 27B | INT4 | 45 tok/s | ~2.7s | |
- | AREX-2 | 27B | INT4 | 45 tok/s | ~2.7s | |
- | PPLX Decider v1 27B | 27B | INT4 | 45 tok/s | ~2.7s | |
- | PPLX Decider v1.1 27B | 27B | INT4 | 45 tok/s | ~2.7s | |
- | Agnes 3.0 Flash | 33B | INT4 | 38 tok/s | ~3.3s | |
- | Falcon-40B | 40B | INT4 | 31 tok/s | ~4.4s | |
Inference Capacity
Maximum model parameter class runnable by weight precision.
Context:
4-bit
32B Models
8-bit
14B Models
16-bit
8B 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 AMD RX 7900 XTX (24GB).
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 960 GB/s aggregate bandwidth, batch size 1 inference operates in a memory-bandwidth bound regime. Generates approximately 213 tok/s on an 8B Q4 model and 25 tok/s on a 70B Q4 model.
Frequently Asked Questions
©2025 ApX Machine Learning