AMD Ryzen AI 9 HX 370 (32GB system, 24GB GPU-accessible)
The AMD Ryzen AI 9 HX 370 (32GB system, 24GB GPU-accessible) provides 24 GB of memory with 90 GB/s bandwidth, supporting local inference via ROCm and Vulkan acceleration.
Memory
24 GB
Usable Memory Ceiling
24 GB
Bandwidth
89.6 GB/s
Max Inference Class
32B Models (Q4)
TDP
54W
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 Ryzen AI 9 HX 370 (32GB system, 24GB GPU-accessible)
Inference throughput, memory compatibility, and generation speeds for open-weights LLMs on AMD Ryzen AI 9 HX 370 (32GB system, 24GB GPU-accessible).
Sort by:
Context:
KV Cache:
GPUs:
| Rank | Model | Parameters | Precision | TPS (Approx.) | TTFT | Can Run |
|---|---|---|---|---|---|---|
#48 | Qwen 3.8 27B | 27B | INT4 | 5 tok/s | ~28.1s | |
#83 | K2 Horizon MoVA 36B A4B | 36B (4B active) | INT4 | 22 tok/s | ~6.0s | |
#93 | Qwen3.5-27B | 27B | INT4 | 5 tok/s | ~28.1s | |
#105 | Sarvam-30B | 32B (2.4B active) | INT4 | 32 tok/s | ~3.8s | |
#106 | Qwen3.6 35B A3B | 35B (3B active) | INT4 | 27 tok/s | ~4.9s | |
#110 | Muse Glimmer 30B | 30B | INT4 | 4 tok/s | ~31.2s | |
#111 | MiMo V2 Flash | 15B (309B active) | INT4 Q8 | 1 tok/s 1 tok/s | ~307.7s ~336.3s | |
#112 | Phi-4 Reasoning Plus | 14B | INT4 Q8 | 6 tok/s 4 tok/s | ~14.8s ~14.9s | |
#117 | Gemma 4 26B A4B | 25.2B (3.8B active) | INT4 | 25 tok/s | ~5.0s | |
#125 | Qwen3.5-9B | 9B | INT4 Q8 FP16 | 14 tok/s 8 tok/s 4 tok/s | ~9.6s ~9.6s ~10.6s | |
#126 | Qwen3.5-35B-A3B | 35B (3B active) | INT4 | 27 tok/s | ~4.9s | |
#128 | K2 Horizon 7B | 7B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
#135 | Gemma 4 12B | 11.95B | INT4 Q8 | 7 tok/s 5 tok/s | ~12.7s ~12.8s | |
#140 | Qwen3.5-4B | 4B | INT4 Q8 FP16 | 29 tok/s 17 tok/s 9 tok/s | ~4.4s ~4.4s ~4.5s | |
#143 | GPT-OSS 20B | 21B (3.6B active) | INT4 | 11 tok/s | ~4.8s | |
#144 | GLM-4V | 9B | INT4 Q8 FP16 | 10 tok/s 6 tok/s 3 tok/s | ~9.6s ~9.7s ~10.6s | |
#150 | NVIDIA Nemotron 3 Nano 30B-A3B | 3.5B (30B active) | INT4 Q8 FP16 | 4 tok/s 2 tok/s 1 tok/s | ~30.2s ~30.4s ~30.8s | |
#153 | Qwen3-30B-A3B | 30B (3B active) | INT4 | 29 tok/s | ~4.4s | |
#158 | Granite 4.2 8B | 8B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
#159 | Gemma 4 E4B | 8B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
#165 | GLM-4 | 32B | INT4 | 3 tok/s | ~39.1s | |
#166 | Phi-4 | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
#167 | Gemma 3 27B | 27B | INT4 | 5 tok/s | ~28.1s | |
#170 | DeepSeek-R1 32B | 32B | INT4 | 4 tok/s | ~33.2s | |
#171 | Qwen3-32B | 32B | INT4 | 4 tok/s | ~33.2s | |
#175 | Magistral Small | 24B | INT4 | 4 tok/s | ~25.1s | |
#176 | OLMo 3.1 32B Instruct | 32B | INT4 | 3 tok/s | ~39.1s | |
#177 | Qwen2.5-32B | 32B | INT4 | 4 tok/s | ~33.2s | |
#178 | OLMo 3 7B Think | 7B | INT4 Q8 FP16 | 12 tok/s 8 tok/s 5 tok/s | ~7.5s ~7.6s ~7.6s | |
#179 | DeepSeek-R1 14B | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
#183 | Ministral 3 14B | 14B | INT4 Q8 | 6 tok/s 4 tok/s | ~14.8s ~14.9s | |
#186 | Gemma 3 12B | 12B | INT4 Q8 FP16 | 11 tok/s 6 tok/s 3 tok/s | ~12.7s ~12.8s ~15.2s | |
#188 | Mistral-Small-2501 | 24B | INT4 | 5 tok/s | ~25.1s | |
#189 | Ministral 3 8B | 8B | INT4 Q8 FP16 | 11 tok/s 7 tok/s 4 tok/s | ~8.6s ~8.6s ~9.4s | |
#190 | Gemma 4 31B | 30.7B | INT4 | 4 tok/s | ~31.9s | |
#191 | MiniCPM5-2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
#192 | Qwen3.5-2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
#193 | Gemma 2 27B | 27B | INT4 | 5 tok/s | ~28.1s | |
#194 | Qwen2.5-14B | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
#195 | OLMo 3.1 32B Think | 32B | INT4 | 3 tok/s | ~39.1s | |
#196 | Nemotron 3.5 Lightning | 30B (3B active) | INT4 | 9 tok/s | ~4.8s | |
#200 | Gemma 4 E2B | 5.1B | INT4 Q8 FP16 | 23 tok/s 14 tok/s 7 tok/s | ~5.5s ~5.6s ~5.6s | |
#202 | DeepSeek-R1 8B | 8B | INT4 Q8 FP16 | 16 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
#204 | Qwen3-4B | 4B | INT4 Q8 FP16 | 29 tok/s 17 tok/s 9 tok/s | ~4.4s ~4.4s ~4.5s | |
#205 | Ministral 3 3B | 3B | INT4 Q8 FP16 | 22 tok/s 16 tok/s 10 tok/s | ~3.4s ~3.4s ~3.4s | |
#206 | OLMo 3 32B Base | 32B | INT4 | 3 tok/s | ~39.1s | |
#207 | Gemma 2 9B | 9B | INT4 Q8 FP16 | 14 tok/s 8 tok/s 4 tok/s | ~9.6s ~9.6s ~10.6s | |
#208 | Qwen3.5-0.8B | 0.8B | INT4 Q8 FP16 | 98 tok/s 67 tok/s 39 tok/s | ~945 ms ~950 ms ~960 ms | |
#209 | Qwen3-14B | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
#210 | Gemma 3 4B | 4B | INT4 Q8 FP16 | 29 tok/s 17 tok/s 9 tok/s | ~4.4s ~4.4s ~4.5s | |
#211 | Qwen3-8B | 8B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
#213 | DeepSeek-R1 7B | 7B | INT4 Q8 FP16 | 18 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
#214 | Qwen3-1.7B | 1.7B | INT4 Q8 FP16 | 56 tok/s 36 tok/s 20 tok/s | ~2.0s ~2.0s ~2.0s | |
#215 | Llama 3.1 8B | 8B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
#216 | Gemma 3 270M | 0.27B | INT4 Q8 FP16 | 136 tok/s 113 tok/s 81 tok/s | ~313 ms ~315 ms ~318 ms | |
#217 | Command R | 35B | INT4 | 3 tok/s | ~42.7s | |
#218 | Qwen3-0.6B | 0.6B | INT4 Q8 FP16 | 118 tok/s 83 tok/s 50 tok/s | ~698 ms ~702 ms ~710 ms | |
#219 | Llama 3 8B | 8B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
#221 | Phi-4-Mini | 3.8B | INT4 Q8 FP16 | 30 tok/s 18 tok/s 9 tok/s | ~4.2s ~4.2s ~4.3s | |
#223 | Qwen2.5-7B | 7B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
#224 | Ministral-8B-2410 | 8B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
#225 | DeepSeek-R1 1.5B | 1.5B | INT4 Q8 FP16 | 62 tok/s 40 tok/s 22 tok/s | ~1.8s ~1.8s ~1.8s | |
#226 | Llama 3.2 3B | 3B | INT4 Q8 FP16 | 36 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
#227 | Nemotron 3.5 Lightning 30B A3B | 30B (3B active) | INT4 | 29 tok/s | ~4.4s | |
#228 | Phi-3-mini | 3.8B | INT4 Q8 FP16 | 30 tok/s 18 tok/s 9 tok/s | ~4.2s ~4.2s ~4.3s | |
#229 | Gemma 3n E2B IT | 6B (2B active) | INT4 Q8 FP16 | 22 tok/s 18 tok/s 12 tok/s | ~2.5s ~2.5s ~2.5s | |
#230 | Mistral-7B-Instruct-v0.1 | 7.3B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.8s ~7.8s ~7.9s | |
#231 | Llama 3.2 1B | 1B | INT4 Q8 FP16 | 83 tok/s 56 tok/s 32 tok/s | ~1.2s ~1.2s ~1.2s | |
#232 | Gemma 2 2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
#233 | Phi-3-medium | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
#234 | OLMo 3 7B Instruct | 7B | INT4 Q8 FP16 | 12 tok/s 8 tok/s 5 tok/s | ~7.5s ~7.6s ~7.6s | |
#235 | Yi-34B | 34B | INT4 | 3 tok/s | ~41.5s | |
#236 | Qwen2-7B | 7B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
#237 | Phi-3-small | 7B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
#238 | Gemma 3 1B | 1B | INT4 Q8 FP16 | 83 tok/s 56 tok/s 32 tok/s | ~1.2s ~1.2s ~1.2s | |
#239 | Mistral-7B-Instruct-v0.2 | 7.3B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.8s ~7.8s ~7.9s | |
#241 | Gemma 1 7B | 7B | INT4 Q8 FP16 | 12 tok/s 8 tok/s 5 tok/s | ~7.5s ~7.6s ~7.6s | |
#242 | Gemma 1 2B | 2B | INT4 Q8 FP16 | 55 tok/s 33 tok/s 18 tok/s | ~2.3s ~2.3s ~2.4s | |
#243 | ChatGLM3-6B | 6B | INT4 Q8 FP16 | 13 tok/s 9 tok/s 5 tok/s | ~6.5s ~6.5s ~6.6s | |
#244 | ChatGLM2-6B | 6B | INT4 Q8 FP16 | 13 tok/s 9 tok/s 5 tok/s | ~6.5s ~6.5s ~6.6s | |
#245 | ChatGLM-6B | 6B | INT4 Q8 FP16 | 13 tok/s 9 tok/s 5 tok/s | ~6.5s ~6.5s ~6.6s | |
Unranked Models (Sorted by Speed) | ||||||
- | DINOv3 ViT-L/16 | 0.3B | INT4 Q8 FP16 | 169 tok/s 130 tok/s 86 tok/s | ~347 ms ~349 ms ~353 ms | |
- | Qwen2-0.5B | 0.5B | INT4 Q8 FP16 | 132 tok/s 95 tok/s 58 tok/s | ~579 ms ~582 ms ~588 ms | |
- | Qwen2.5-0.5B | 0.5B | INT4 Q8 FP16 | 132 tok/s 95 tok/s 58 tok/s | ~579 ms ~582 ms ~588 ms | |
- | ERNIE-4.5-0.3B | 0.3B | INT4 Q8 FP16 | 126 tok/s 105 tok/s 74 tok/s | ~349 ms ~350 ms ~354 ms | |
- | ERNIE-4.5-0.3B-Base | 0.3B | INT4 Q8 FP16 | 126 tok/s 105 tok/s 74 tok/s | ~349 ms ~350 ms ~354 ms | |
- | Liquid D1-Omni 600M | 0.6B | INT4 Q8 FP16 | 118 tok/s 83 tok/s 50 tok/s | ~698 ms ~702 ms ~710 ms | |
- | PixArt-α XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 117 tok/s 82 tok/s 50 tok/s | ~709 ms ~713 ms ~721 ms | |
- | PixArt-Σ XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 117 tok/s 82 tok/s 50 tok/s | ~709 ms ~713 ms ~721 ms | |
- | Nemotron 3.5 Parse | 0.63B | INT4 Q8 FP16 | 115 tok/s 80 tok/s 48 tok/s | ~737 ms ~741 ms ~749 ms | |
- | SD 2.1 Turbo | 0.86B | INT4 Q8 FP16 | 93 tok/s 63 tok/s 37 tok/s | ~1.0s ~1.0s ~1.0s | |
- | Stable Diffusion 1.4 | 0.86B | INT4 Q8 FP16 | 93 tok/s 63 tok/s 37 tok/s | ~1.0s ~1.0s ~1.0s | |
- | Stable Diffusion 2.1 | 0.86B | INT4 Q8 FP16 | 93 tok/s 63 tok/s 37 tok/s | ~1.0s ~1.0s ~1.0s | |
- | Falcon-1B | 1B | INT4 Q8 FP16 | 91 tok/s 59 tok/s 33 tok/s | ~1.2s ~1.2s ~1.2s | |
- | Falcon3-1B | 1B | INT4 Q8 FP16 | 83 tok/s 56 tok/s 32 tok/s | ~1.2s ~1.2s ~1.2s | |
- | MiniCPM-V 4.7 1B | 1B | INT4 Q8 FP16 | 83 tok/s 56 tok/s 32 tok/s | ~1.2s ~1.2s ~1.2s | |
- | Kandinsky 2.2 Decoder | 1.22B | INT4 Q8 FP16 | 72 tok/s 48 tok/s 27 tok/s | ~1.4s ~1.5s ~1.5s | |
- | Wan 2.1 T2V 1.3B | 1.3B | INT4 Q8 FP16 | 69 tok/s 45 tok/s 26 tok/s | ~1.5s ~1.5s ~1.6s | |
- | Qwen2-1.5B | 1.5B | INT4 Q8 FP16 | 62 tok/s 40 tok/s 22 tok/s | ~1.8s ~1.8s ~1.8s | |
- | Qwen2.5-1.5B | 1.5B | INT4 Q8 FP16 | 62 tok/s 40 tok/s 22 tok/s | ~1.8s ~1.8s ~1.8s | |
- | SVD Image-to-Video XT | 1.5B | INT4 Q8 FP16 | 62 tok/s 40 tok/s 22 tok/s | ~1.8s ~1.8s ~1.8s | |
- | SVD Image-to-Video XT 1.1 | 1.5B | INT4 Q8 FP16 | 62 tok/s 40 tok/s 22 tok/s | ~1.8s ~1.8s ~1.8s | |
- | LFM3 1.7B | 1.7B | INT4 Q8 FP16 | 56 tok/s 36 tok/s 20 tok/s | ~2.0s ~2.0s ~2.0s | |
- | LFM3 1.7B Base | 1.7B | INT4 Q8 FP16 | 56 tok/s 36 tok/s 20 tok/s | ~2.0s ~2.0s ~2.0s | |
- | Spark X2.5 1.7B | 1.7B | INT4 Q8 FP16 | 56 tok/s 36 tok/s 20 tok/s | ~2.0s ~2.0s ~2.0s | |
- | Hy-MT2 1.8B | 1.8B | INT4 Q8 FP16 | 54 tok/s 34 tok/s 19 tok/s | ~2.1s ~2.1s ~2.1s | |
- | CogVideoX 2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
- | LFM3 Audio 2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
- | LFM3 Omni 2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
- | LFM3 VL 2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
- | Liquid D1 2B | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
- | LTX-Video 0.9.1 | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
- | Moondream 3.1 | 9B (2B active) | INT4 Q8 FP16 | 50 tok/s 31 tok/s 16 tok/s | ~2.6s ~2.6s ~2.9s | |
- | Stable Diffusion 3 Medium | 2B | INT4 Q8 FP16 | 50 tok/s 31 tok/s 17 tok/s | ~2.3s ~2.3s ~2.4s | |
- | Kandinsky 3.0 | 2.1B | INT4 Q8 FP16 | 48 tok/s 30 tok/s 16 tok/s | ~2.4s ~2.4s ~2.5s | |
- | LTX-2 | 2.2B | INT4 Q8 FP16 | 46 tok/s 29 tok/s 16 tok/s | ~2.5s ~2.5s ~2.6s | |
- | SDXL 1.0 Refiner | 2.31B | INT4 Q8 FP16 | 45 tok/s 28 tok/s 15 tok/s | ~2.6s ~2.7s ~2.7s | |
- | Stable Diffusion 3.5 Medium | 2.5B | INT4 Q8 FP16 | 42 tok/s 26 tok/s 14 tok/s | ~2.8s ~2.9s ~2.9s | |
- | SDXL 1.0 Base | 2.56B | INT4 Q8 FP16 | 41 tok/s 25 tok/s 14 tok/s | ~2.9s ~2.9s ~3.0s | |
- | SDXL Turbo | 2.56B | INT4 Q8 FP16 | 41 tok/s 25 tok/s 14 tok/s | ~2.9s ~2.9s ~3.0s | |
- | Falcon-3B | 3B | INT4 Q8 FP16 | 39 tok/s 23 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | CroissantLLM Base | 1.3B | INT4 Q8 FP16 | 38 tok/s 30 tok/s 20 tok/s | ~1.5s ~1.6s ~1.6s | |
- | Phi-1 | 1.3B | INT4 Q8 FP16 | 38 tok/s 30 tok/s 20 tok/s | ~1.5s ~1.6s ~1.6s | |
- | Phi-1.5 | 1.3B | INT4 Q8 FP16 | 38 tok/s 30 tok/s 20 tok/s | ~1.5s ~1.6s ~1.6s | |
- | DeepSeek-R1 3B | 3B | INT4 Q8 FP16 | 37 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | Mellum 2.1 12B A2.5B Thinking | 12B (2.5B active) | INT4 Q8 FP16 | 37 tok/s 24 tok/s 12 tok/s | ~3.3s ~3.3s ~3.9s | |
- | Falcon3-3B | 3B | INT4 Q8 FP16 | 36 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | LFM2-VL 3B | 3B | INT4 Q8 FP16 | 36 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | Liquid D1 3B | 3B | INT4 Q8 FP16 | 36 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | Liquid D1-VL 3B | 3B | INT4 Q8 FP16 | 36 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | Liquid LFM2-VL 3B D1 | 3B | INT4 Q8 FP16 | 36 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | Qwen2.5-3B | 3B | INT4 Q8 FP16 | 36 tok/s 22 tok/s 12 tok/s | ~3.4s ~3.4s ~3.4s | |
- | LFM2.5 24B A2B | 24B (2.5B active) | INT4 | 33 tok/s | ~3.7s | |
- | ERNIE-4.5-21B-A3B | 21B (3B active) | INT4 Q8 | 30 tok/s 18 tok/s | ~4.1s ~4.8s | |
- | ERNIE-4.5-21B-A3B-Base | 21B (3B active) | INT4 Q8 | 30 tok/s 18 tok/s | ~4.1s ~4.8s | |
- | ERNIE-4.5-VL-28B-A3B | 28B (3B active) | INT4 | 29 tok/s | ~4.3s | |
- | ERNIE-4.5-VL-28B-A3B-Base | 28B (3B active) | INT4 | 29 tok/s | ~4.3s | |
- | FLUX.2 Klein 4B | 4B | INT4 Q8 FP16 | 29 tok/s 17 tok/s 9 tok/s | ~4.4s ~4.4s ~4.5s | |
- | FrogNano 4B | 4B | INT4 Q8 FP16 | 29 tok/s 17 tok/s 9 tok/s | ~4.4s ~4.4s ~4.5s | |
- | Hy-Memory 30B A3B | 30B (3B active) | INT4 | 29 tok/s | ~4.4s | |
- | Hy-MT2 30B A3B | 30B (3B active) | INT4 | 29 tok/s | ~4.4s | |
- | Mage-VL Cowork | 4B | INT4 Q8 FP16 | 29 tok/s 17 tok/s 9 tok/s | ~4.4s ~4.4s ~4.5s | |
- | Spark X2.5 4B | 4B | INT4 Q8 FP16 | 29 tok/s 17 tok/s 9 tok/s | ~4.4s ~4.4s ~4.5s | |
- | EmbeddingGemma 2 | 2B | INT4 Q8 FP16 | 27 tok/s 21 tok/s 14 tok/s | ~2.3s ~2.4s ~2.4s | |
- | MiniCPM-V 4.7 35B A3B | 35B (3B active) | INT4 | 27 tok/s | ~4.9s | |
- | CogVideoX 1.5 5B | 5B | INT4 Q8 FP16 | 24 tok/s 14 tok/s 7 tok/s | ~5.4s ~5.5s ~5.5s | |
- | CogVideoX 5B | 5B | INT4 Q8 FP16 | 24 tok/s 14 tok/s 7 tok/s | ~5.4s ~5.5s ~5.5s | |
- | Wan 2.2 TI2V 5B | 5B | INT4 Q8 FP16 | 24 tok/s 14 tok/s 7 tok/s | ~5.4s ~5.5s ~5.5s | |
- | Phi-2 | 2.7B | INT4 Q8 FP16 | 23 tok/s 17 tok/s 11 tok/s | ~3.1s ~3.1s ~3.1s | |
- | MaLLaM-3B | 3B | INT4 Q8 FP16 | 22 tok/s 16 tok/s 10 tok/s | ~3.4s ~3.4s ~3.4s | |
- | SmolLM3 3B | 3B | INT4 Q8 FP16 | 22 tok/s 16 tok/s 10 tok/s | ~3.4s ~3.4s ~3.4s | |
- | AuraFlow v0.2 | 6.8B | INT4 Q8 FP16 | 18 tok/s 10 tok/s 5 tok/s | ~7.3s ~7.3s ~7.4s | |
- | AuraFlow v0.3 | 6.8B | INT4 Q8 FP16 | 18 tok/s 10 tok/s 5 tok/s | ~7.3s ~7.3s ~7.4s | |
- | Falcon-7B | 7B | INT4 Q8 FP16 | 18 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
- | Falcon3-7B | 7B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
- | Hy-MT2 7B | 7B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.5s ~7.5s ~7.6s | |
- | Mistral-7B-v0.1 | 7.3B | INT4 Q8 FP16 | 17 tok/s 10 tok/s 5 tok/s | ~7.8s ~7.8s ~7.9s | |
- | Youtu Parsing Omni | 8B | INT4 Q8 FP16 | 16 tok/s 9 tok/s 5 tok/s | ~8.5s ~8.6s ~8.7s | |
- | Stable Diffusion 3.5 Large | 8.1B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.6s ~8.7s ~8.8s | |
- | Stable Diffusion 3.5 Large Turbo | 8.1B | INT4 Q8 FP16 | 15 tok/s 9 tok/s 5 tok/s | ~8.6s ~8.7s ~8.8s | |
- | FLUX.2 Klein 9B | 9B | INT4 Q8 FP16 | 14 tok/s 8 tok/s 4 tok/s | ~9.6s ~9.6s ~10.6s | |
- | LensVLM 9B | 9B | INT4 Q8 FP16 | 14 tok/s 8 tok/s 4 tok/s | ~9.6s ~9.6s ~10.6s | |
- | MiMo V2.6 Distill Qwen 9B | 9B | INT4 Q8 FP16 | 14 tok/s 8 tok/s 4 tok/s | ~9.6s ~9.6s ~10.6s | |
- | ChatGLM3-6B-32K | 6B | INT4 Q8 FP16 | 13 tok/s 9 tok/s 5 tok/s | ~6.5s ~6.5s ~6.6s | |
- | Falcon3-10B | 10B | INT4 Q8 FP16 | 13 tok/s 7 tok/s 4 tok/s | ~10.6s ~10.7s ~11.7s | |
- | Kimi-VL-A3B-Instruct | 16B (3B active) | INT4 Q8 | 13 tok/s 10 tok/s | ~4.0s ~4.3s | |
- | Mochi 1 Preview | 10B | INT4 Q8 FP16 | 13 tok/s 7 tok/s 4 tok/s | ~10.6s ~10.7s ~11.7s | |
- | Yi-6B | 6B | INT4 Q8 FP16 | 13 tok/s 9 tok/s 5 tok/s | ~6.5s ~6.5s ~6.6s | |
- | Falcon2-11B | 11B | INT4 Q8 FP16 | 12 tok/s 7 tok/s 3 tok/s | ~11.6s ~11.7s ~13.9s | |
- | Hunyuan Lite | 7B | INT4 Q8 FP16 | 12 tok/s 8 tok/s 5 tok/s | ~7.5s ~7.6s ~7.6s | |
- | OLMo 3 7B Base | 7B | INT4 Q8 FP16 | 12 tok/s 8 tok/s 5 tok/s | ~7.5s ~7.6s ~7.6s | |
- | SEA-LION-7B | 7.1B | INT4 Q8 FP16 | 12 tok/s 8 tok/s 5 tok/s | ~7.6s ~7.7s ~7.8s | |
- | SEA-LION-7B-Instruct | 7.1B | INT4 Q8 FP16 | 12 tok/s 8 tok/s 5 tok/s | ~7.6s ~7.7s ~7.8s | |
- | FLUX.1 Canny Dev | 11.9B | INT4 Q8 FP16 | 11 tok/s 6 tok/s 3 tok/s | ~12.6s ~12.7s ~15.1s | |
- | FLUX.1 Depth Dev | 11.9B | INT4 Q8 FP16 | 11 tok/s 6 tok/s 3 tok/s | ~12.6s ~12.7s ~15.1s | |
- | FLUX.1 Dev | 11.9B | INT4 Q8 FP16 | 11 tok/s 6 tok/s 3 tok/s | ~12.6s ~12.7s ~15.1s | |
- | FLUX.1 Fill Dev | 11.9B | INT4 Q8 FP16 | 11 tok/s 6 tok/s 3 tok/s | ~12.6s ~12.7s ~15.1s | |
- | FLUX.1 Schnell | 11.9B | INT4 Q8 FP16 | 11 tok/s 6 tok/s 3 tok/s | ~12.6s ~12.7s ~15.1s | |
- | Sahabat-AI-Llama3-8B-Instruct | 8B | INT4 Q8 FP16 | 11 tok/s 7 tok/s 4 tok/s | ~8.6s ~8.6s ~9.4s | |
- | Typhoon-2-8B | 8B | INT4 Q8 FP16 | 11 tok/s 7 tok/s 4 tok/s | ~8.6s ~8.6s ~9.4s | |
- | GLM-4-9B | 9B | INT4 Q8 FP16 | 10 tok/s 6 tok/s 3 tok/s | ~9.6s ~9.7s ~10.6s | |
- | GLM-4-9B-Chat | 9B | INT4 Q8 FP16 | 10 tok/s 6 tok/s 3 tok/s | ~9.6s ~9.7s ~10.6s | |
- | GLM-4-9B-Chat-1M | 9B | INT4 Q8 FP16 | 10 tok/s 6 tok/s 3 tok/s | ~9.6s ~9.7s ~10.6s | |
- | HunyuanVideo 1.5 720P T2V | 13B | INT4 Q8 | 10 tok/s 6 tok/s | ~13.7s ~13.8s | |
- | Yi-9B | 9B | INT4 Q8 FP16 | 10 tok/s 6 tok/s 3 tok/s | ~9.6s ~9.7s ~10.6s | |
- | FLUX.2 Dev | 15B | INT4 Q8 | 9 tok/s 5 tok/s | ~15.8s ~15.9s | |
- | Sahabat-AI-Gemma2-9B | 9.2B | INT4 Q8 FP16 | 9 tok/s 6 tok/s 3 tok/s | ~9.8s ~9.9s ~10.8s | |
- | Sahabat-AI-Gemma2-9B-Instruct | 9.2B | INT4 Q8 FP16 | 9 tok/s 6 tok/s 3 tok/s | ~9.8s ~9.9s ~10.8s | |
- | Wan 2.1 I2V 14B 720P | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
- | Wan 2.1 T2V 14B | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
- | Wan 2.2 T2V A14B | 14B | INT4 Q8 | 9 tok/s 5 tok/s | ~14.8s ~14.9s | |
- | GreenMind-14B-R1 | 14B | INT4 Q8 | 6 tok/s 4 tok/s | ~14.8s ~14.9s | |
- | Arete 27B | 27B | INT4 | 5 tok/s | ~28.1s | |
- | AREX-2 | 27B | INT4 | 5 tok/s | ~28.1s | |
- | PPLX Decider v1 27B | 27B | INT4 | 5 tok/s | ~28.1s | |
- | PPLX Decider v1.1 27B | 27B | INT4 | 5 tok/s | ~28.1s | |
- | Agnes 3.0 Flash | 33B | INT4 | 4 tok/s | ~34.2s | |
- | Falcon-40B | 40B | INT4 | 3 tok/s | ~44.9s | |
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 Ryzen AI 9 HX 370 (32GB system, 24GB GPU-accessible).
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 89.6 GB/s aggregate bandwidth, batch size 1 inference operates in a memory-bandwidth bound regime. Generates approximately 20 tok/s on an 8B Q4 model and 2 tok/s on a 70B Q4 model.
Frequently Asked Questions
©2025 ApX Machine Learning