NVIDIA RTX A4000 (16GB)
The NVIDIA RTX A4000 (16GB) features 16 GB of dedicated VRAM and 448 GB/s memory bandwidth, supporting CUDA and TensorRT-LLM runtimes for full-precision and quantized inference.
Memory
16 GB
Usable Memory Ceiling
16 GB
Bandwidth
448 GB/s
Max Inference Class
14B Models (Q4)
TDP
140W
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 NVIDIA RTX A4000 (16GB)
Inference throughput, memory compatibility, and generation speeds for open-weights LLMs on NVIDIA RTX A4000 (16GB).
Sort by:
Context:
KV Cache:
GPUs:
| Rank | Model | Parameters | Precision | TPS (Approx.) | TTFT | Can Run |
|---|---|---|---|---|---|---|
#48 | Qwen 3.8 27B | 27B | INT4 | 21 tok/s | ~4.5s | |
#93 | Qwen3.5-27B | 27B | INT4 | 21 tok/s | ~4.5s | |
#111 | MiMo V2 Flash | 15B (309B active) | INT4 | 2 tok/s | ~45.6s | |
#112 | Phi-4 Reasoning Plus | 14B | INT4 | 31 tok/s | ~2.0s | |
#117 | Gemma 4 26B A4B | 25.2B (3.8B active) | INT4 | 96 tok/s | ~750 ms | |
#125 | Qwen3.5-9B | 9B | INT4 Q8 | 61 tok/s 37 tok/s | ~1.3s ~1.3s | |
#128 | K2 Horizon 7B | 7B | INT4 Q8 FP16 | 74 tok/s 46 tok/s 22 tok/s | ~1.0s ~1.0s ~1.2s | |
#135 | Gemma 4 12B | 11.95B | INT4 Q8 | 35 tok/s 21 tok/s | ~1.7s ~2.1s | |
#140 | Qwen3.5-4B | 4B | INT4 Q8 FP16 | 111 tok/s 73 tok/s 41 tok/s | ~604 ms ~609 ms ~619 ms | |
#143 | GPT-OSS 20B | 21B (3.6B active) | INT4 | 43 tok/s | ~771 ms | |
#144 | GLM-4V | 9B | INT4 Q8 | 44 tok/s 29 tok/s | ~1.3s ~1.4s | |
#150 | NVIDIA Nemotron 3 Nano 30B-A3B | 3.5B (30B active) | INT4 Q8 FP16 | 20 tok/s 12 tok/s 6 tok/s | ~4.1s ~4.2s ~4.2s | |
#158 | Granite 4.2 8B | 8B | INT4 Q8 | 67 tok/s 41 tok/s | ~1.2s ~1.2s | |
#159 | Gemma 4 E4B | 8B | INT4 Q8 | 67 tok/s 41 tok/s | ~1.2s ~1.2s | |
#166 | Phi-4 | 14B | INT4 Q8 | 42 tok/s 22 tok/s | ~2.0s ~2.4s | |
#167 | Gemma 3 27B | 27B | INT4 | 21 tok/s | ~4.5s | |
#178 | OLMo 3 7B Think | 7B | INT4 Q8 | 53 tok/s 37 tok/s | ~1.0s ~1.0s | |
#179 | DeepSeek-R1 14B | 14B | INT4 Q8 | 43 tok/s 23 tok/s | ~2.0s ~2.4s | |
#183 | Ministral 3 14B | 14B | INT4 | 31 tok/s | ~2.0s | |
#186 | Gemma 3 12B | 12B | INT4 Q8 | 48 tok/s 27 tok/s | ~1.7s ~1.9s | |
#188 | Mistral-Small-2501 | 24B | INT4 | 25 tok/s | ~3.7s | |
#189 | Ministral 3 8B | 8B | INT4 Q8 | 48 tok/s 34 tok/s | ~1.2s ~1.2s | |
#191 | MiniCPM5-2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
#192 | Qwen3.5-2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
#193 | Gemma 2 27B | 27B | INT4 | 21 tok/s | ~4.5s | |
#194 | Qwen2.5-14B | 14B | INT4 Q8 | 42 tok/s 22 tok/s | ~2.0s ~2.4s | |
#200 | Gemma 4 E2B | 5.1B | INT4 Q8 FP16 | 94 tok/s 60 tok/s 33 tok/s | ~759 ms ~765 ms ~779 ms | |
#202 | DeepSeek-R1 8B | 8B | INT4 Q8 | 69 tok/s 41 tok/s | ~1.2s ~1.2s | |
#204 | Qwen3-4B | 4B | INT4 Q8 FP16 | 111 tok/s 73 tok/s 41 tok/s | ~604 ms ~609 ms ~619 ms | |
#205 | Ministral 3 3B | 3B | INT4 Q8 FP16 | 89 tok/s 69 tok/s 45 tok/s | ~467 ms ~470 ms ~478 ms | |
#207 | Gemma 2 9B | 9B | INT4 Q8 | 61 tok/s 37 tok/s | ~1.3s ~1.3s | |
#208 | Qwen3.5-0.8B | 0.8B | INT4 Q8 FP16 | 247 tok/s 200 tok/s 141 tok/s | ~135 ms ~136 ms ~138 ms | |
#209 | Qwen3-14B | 14B | INT4 Q8 | 42 tok/s 22 tok/s | ~2.0s ~2.4s | |
#210 | Gemma 3 4B | 4B | INT4 Q8 FP16 | 111 tok/s 73 tok/s 41 tok/s | ~604 ms ~609 ms ~619 ms | |
#211 | Qwen3-8B | 8B | INT4 Q8 | 67 tok/s 41 tok/s | ~1.2s ~1.2s | |
#213 | DeepSeek-R1 7B | 7B | INT4 Q8 FP16 | 76 tok/s 47 tok/s 24 tok/s | ~1.0s ~1.0s ~1.1s | |
#214 | Qwen3-1.7B | 1.7B | INT4 Q8 FP16 | 180 tok/s 132 tok/s 83 tok/s | ~277 ms ~280 ms ~284 ms | |
#215 | Llama 3.1 8B | 8B | INT4 Q8 | 67 tok/s 41 tok/s | ~1.2s ~1.2s | |
#216 | Gemma 3 270M | 0.27B | INT4 Q8 FP16 | 288 tok/s 265 tok/s 223 tok/s | ~49 ms ~49 ms ~50 ms | |
#218 | Qwen3-0.6B | 0.6B | INT4 Q8 FP16 | 271 tok/s 227 tok/s 167 tok/s | ~102 ms ~102 ms ~104 ms | |
#219 | Llama 3 8B | 8B | INT4 Q8 | 67 tok/s 41 tok/s | ~1.2s ~1.2s | |
#221 | Phi-4-Mini | 3.8B | INT4 Q8 FP16 | 115 tok/s 76 tok/s 43 tok/s | ~576 ms ~580 ms ~590 ms | |
#223 | Qwen2.5-7B | 7B | INT4 Q8 FP16 | 74 tok/s 46 tok/s 22 tok/s | ~1.0s ~1.0s ~1.2s | |
#224 | Ministral-8B-2410 | 8B | INT4 Q8 | 67 tok/s 41 tok/s | ~1.2s ~1.2s | |
#225 | DeepSeek-R1 1.5B | 1.5B | INT4 Q8 FP16 | 192 tok/s 143 tok/s 92 tok/s | ~246 ms ~248 ms ~252 ms | |
#226 | Llama 3.2 3B | 3B | INT4 Q8 FP16 | 133 tok/s 90 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
#228 | Phi-3-mini | 3.8B | INT4 Q8 FP16 | 115 tok/s 76 tok/s 43 tok/s | ~576 ms ~580 ms ~590 ms | |
#229 | Gemma 3n E2B IT | 6B (2B active) | INT4 Q8 FP16 | 89 tok/s 76 tok/s 52 tok/s | ~350 ms ~352 ms ~387 ms | |
#230 | Mistral-7B-Instruct-v0.1 | 7.3B | INT4 Q8 FP16 | 72 tok/s 44 tok/s 22 tok/s | ~1.1s ~1.1s ~1.3s | |
#231 | Llama 3.2 1B | 1B | INT4 Q8 FP16 | 227 tok/s 179 tok/s 122 tok/s | ~170 ms ~171 ms ~174 ms | |
#232 | Gemma 2 2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
#233 | Phi-3-medium | 14B | INT4 Q8 | 42 tok/s 22 tok/s | ~2.0s ~2.4s | |
#234 | OLMo 3 7B Instruct | 7B | INT4 Q8 | 53 tok/s 37 tok/s | ~1.0s ~1.0s | |
#236 | Qwen2-7B | 7B | INT4 Q8 FP16 | 74 tok/s 46 tok/s 22 tok/s | ~1.0s ~1.0s ~1.2s | |
#237 | Phi-3-small | 7B | INT4 Q8 FP16 | 74 tok/s 46 tok/s 22 tok/s | ~1.0s ~1.0s ~1.2s | |
#238 | Gemma 3 1B | 1B | INT4 Q8 FP16 | 227 tok/s 179 tok/s 122 tok/s | ~170 ms ~171 ms ~174 ms | |
#239 | Mistral-7B-Instruct-v0.2 | 7.3B | INT4 Q8 FP16 | 72 tok/s 44 tok/s 22 tok/s | ~1.1s ~1.1s ~1.3s | |
#241 | Gemma 1 7B | 7B | INT4 Q8 | 53 tok/s 37 tok/s | ~1.0s ~1.0s | |
#242 | Gemma 1 2B | 2B | INT4 Q8 FP16 | 176 tok/s 124 tok/s 75 tok/s | ~323 ms ~326 ms ~331 ms | |
#243 | ChatGLM3-6B | 6B | INT4 Q8 FP16 | 59 tok/s 42 tok/s 24 tok/s | ~891 ms ~898 ms ~991 ms | |
#244 | ChatGLM2-6B | 6B | INT4 Q8 FP16 | 59 tok/s 42 tok/s 24 tok/s | ~891 ms ~898 ms ~991 ms | |
#245 | ChatGLM-6B | 6B | INT4 Q8 FP16 | 59 tok/s 42 tok/s 24 tok/s | ~891 ms ~898 ms ~991 ms | |
Unranked Models (Sorted by Speed) | ||||||
- | DINOv3 ViT-L/16 | 0.3B | INT4 Q8 FP16 | 314 tok/s 282 tok/s 231 tok/s | ~54 ms ~54 ms ~55 ms | |
- | Qwen2-0.5B | 0.5B | INT4 Q8 FP16 | 284 tok/s 243 tok/s 184 tok/s | ~85 ms ~86 ms ~87 ms | |
- | Qwen2.5-0.5B | 0.5B | INT4 Q8 FP16 | 284 tok/s 243 tok/s 184 tok/s | ~85 ms ~86 ms ~87 ms | |
- | ERNIE-4.5-0.3B | 0.3B | INT4 Q8 FP16 | 279 tok/s 256 tok/s 213 tok/s | ~54 ms ~54 ms ~55 ms | |
- | ERNIE-4.5-0.3B-Base | 0.3B | INT4 Q8 FP16 | 279 tok/s 256 tok/s 213 tok/s | ~54 ms ~54 ms ~55 ms | |
- | Liquid D1-Omni 600M | 0.6B | INT4 Q8 FP16 | 271 tok/s 227 tok/s 167 tok/s | ~102 ms ~102 ms ~104 ms | |
- | PixArt-α XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 270 tok/s 226 tok/s 166 tok/s | ~103 ms ~104 ms ~105 ms | |
- | PixArt-Σ XL 2 1024-MS | 0.61B | INT4 Q8 FP16 | 270 tok/s 226 tok/s 166 tok/s | ~103 ms ~104 ms ~105 ms | |
- | Nemotron 3.5 Parse | 0.63B | INT4 Q8 FP16 | 267 tok/s 223 tok/s 163 tok/s | ~107 ms ~108 ms ~109 ms | |
- | SD 2.1 Turbo | 0.86B | INT4 Q8 FP16 | 241 tok/s 194 tok/s 135 tok/s | ~145 ms ~146 ms ~149 ms | |
- | Stable Diffusion 1.4 | 0.86B | INT4 Q8 FP16 | 241 tok/s 194 tok/s 135 tok/s | ~145 ms ~146 ms ~149 ms | |
- | Stable Diffusion 2.1 | 0.86B | INT4 Q8 FP16 | 241 tok/s 194 tok/s 135 tok/s | ~145 ms ~146 ms ~149 ms | |
- | Falcon-1B | 1B | INT4 Q8 FP16 | 238 tok/s 186 tok/s 125 tok/s | ~170 ms ~171 ms ~174 ms | |
- | Falcon3-1B | 1B | INT4 Q8 FP16 | 227 tok/s 179 tok/s 122 tok/s | ~170 ms ~171 ms ~174 ms | |
- | MiniCPM-V 4.7 1B | 1B | INT4 Q8 FP16 | 227 tok/s 179 tok/s 122 tok/s | ~170 ms ~171 ms ~174 ms | |
- | Kandinsky 2.2 Decoder | 1.22B | INT4 Q8 FP16 | 210 tok/s 161 tok/s 106 tok/s | ~204 ms ~205 ms ~208 ms | |
- | Wan 2.1 T2V 1.3B | 1.3B | INT4 Q8 FP16 | 204 tok/s 156 tok/s 102 tok/s | ~215 ms ~217 ms ~220 ms | |
- | Qwen2-1.5B | 1.5B | INT4 Q8 FP16 | 192 tok/s 143 tok/s 92 tok/s | ~246 ms ~248 ms ~252 ms | |
- | Qwen2.5-1.5B | 1.5B | INT4 Q8 FP16 | 192 tok/s 143 tok/s 92 tok/s | ~246 ms ~248 ms ~252 ms | |
- | SVD Image-to-Video XT | 1.5B | INT4 Q8 FP16 | 192 tok/s 143 tok/s 92 tok/s | ~246 ms ~248 ms ~252 ms | |
- | SVD Image-to-Video XT 1.1 | 1.5B | INT4 Q8 FP16 | 192 tok/s 143 tok/s 92 tok/s | ~246 ms ~248 ms ~252 ms | |
- | LFM3 1.7B | 1.7B | INT4 Q8 FP16 | 180 tok/s 132 tok/s 83 tok/s | ~277 ms ~280 ms ~284 ms | |
- | LFM3 1.7B Base | 1.7B | INT4 Q8 FP16 | 180 tok/s 132 tok/s 83 tok/s | ~277 ms ~280 ms ~284 ms | |
- | Spark X2.5 1.7B | 1.7B | INT4 Q8 FP16 | 180 tok/s 132 tok/s 83 tok/s | ~277 ms ~280 ms ~284 ms | |
- | Hy-MT2 1.8B | 1.8B | INT4 Q8 FP16 | 175 tok/s 128 tok/s 80 tok/s | ~292 ms ~294 ms ~299 ms | |
- | CogVideoX 2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
- | LFM3 Audio 2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
- | LFM3 Omni 2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
- | LFM3 VL 2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
- | Liquid D1 2B | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
- | LTX-Video 0.9.1 | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
- | Moondream 3.1 | 9B (2B active) | INT4 Q8 | 166 tok/s 119 tok/s | ~363 ms ~366 ms | |
- | Stable Diffusion 3 Medium | 2B | INT4 Q8 FP16 | 166 tok/s 119 tok/s 73 tok/s | ~323 ms ~326 ms ~331 ms | |
- | Kandinsky 3.0 | 2.1B | INT4 Q8 FP16 | 162 tok/s 115 tok/s 71 tok/s | ~337 ms ~340 ms ~345 ms | |
- | LTX-2 | 2.2B | INT4 Q8 FP16 | 158 tok/s 112 tok/s 68 tok/s | ~351 ms ~354 ms ~360 ms | |
- | SDXL 1.0 Refiner | 2.31B | INT4 Q8 FP16 | 154 tok/s 108 tok/s 65 tok/s | ~367 ms ~369 ms ~376 ms | |
- | Stable Diffusion 3.5 Medium | 2.5B | INT4 Q8 FP16 | 148 tok/s 103 tok/s 61 tok/s | ~393 ms ~396 ms ~403 ms | |
- | SDXL 1.0 Base | 2.56B | INT4 Q8 FP16 | 146 tok/s 101 tok/s 60 tok/s | ~402 ms ~405 ms ~411 ms | |
- | SDXL Turbo | 2.56B | INT4 Q8 FP16 | 146 tok/s 101 tok/s 60 tok/s | ~402 ms ~405 ms ~411 ms | |
- | Falcon-3B | 3B | INT4 Q8 FP16 | 141 tok/s 94 tok/s 54 tok/s | ~463 ms ~466 ms ~474 ms | |
- | CroissantLLM Base | 1.3B | INT4 Q8 FP16 | 138 tok/s 117 tok/s 83 tok/s | ~217 ms ~219 ms ~222 ms | |
- | Phi-1 | 1.3B | INT4 Q8 FP16 | 138 tok/s 117 tok/s 83 tok/s | ~217 ms ~219 ms ~222 ms | |
- | Phi-1.5 | 1.3B | INT4 Q8 FP16 | 138 tok/s 117 tok/s 83 tok/s | ~217 ms ~219 ms ~222 ms | |
- | DeepSeek-R1 3B | 3B | INT4 Q8 FP16 | 135 tok/s 91 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
- | Mellum 2.1 12B A2.5B Thinking | 12B (2.5B active) | INT4 Q8 | 135 tok/s 92 tok/s | ~452 ms ~494 ms | |
- | Falcon3-3B | 3B | INT4 Q8 FP16 | 133 tok/s 90 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
- | LFM2-VL 3B | 3B | INT4 Q8 FP16 | 133 tok/s 90 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
- | Liquid D1 3B | 3B | INT4 Q8 FP16 | 133 tok/s 90 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
- | Liquid D1-VL 3B | 3B | INT4 Q8 FP16 | 133 tok/s 90 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
- | Liquid LFM2-VL 3B D1 | 3B | INT4 Q8 FP16 | 133 tok/s 90 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
- | Qwen2.5-3B | 3B | INT4 Q8 FP16 | 133 tok/s 90 tok/s 53 tok/s | ~463 ms ~467 ms ~475 ms | |
- | LFM2.5 24B A2B | 24B (2.5B active) | INT4 | 118 tok/s | ~556 ms | |
- | ERNIE-4.5-21B-A3B | 21B (3B active) | INT4 | 111 tok/s | ~613 ms | |
- | ERNIE-4.5-21B-A3B-Base | 21B (3B active) | INT4 | 111 tok/s | ~613 ms | |
- | FLUX.2 Klein 4B | 4B | INT4 Q8 FP16 | 111 tok/s 73 tok/s 41 tok/s | ~604 ms ~609 ms ~619 ms | |
- | FrogNano 4B | 4B | INT4 Q8 FP16 | 111 tok/s 73 tok/s 41 tok/s | ~604 ms ~609 ms ~619 ms | |
- | Mage-VL Cowork | 4B | INT4 Q8 FP16 | 111 tok/s 73 tok/s 41 tok/s | ~604 ms ~609 ms ~619 ms | |
- | Spark X2.5 4B | 4B | INT4 Q8 FP16 | 111 tok/s 73 tok/s 41 tok/s | ~604 ms ~609 ms ~619 ms | |
- | EmbeddingGemma 2 | 2B | INT4 Q8 FP16 | 107 tok/s 87 tok/s 60 tok/s | ~327 ms ~329 ms ~334 ms | |
- | ERNIE-4.5-VL-28B-A3B | 28B (3B active) | INT4 | 101 tok/s | ~697 ms | |
- | ERNIE-4.5-VL-28B-A3B-Base | 28B (3B active) | INT4 | 101 tok/s | ~697 ms | |
- | CogVideoX 1.5 5B | 5B | INT4 Q8 FP16 | 95 tok/s 61 tok/s 34 tok/s | ~745 ms ~751 ms ~764 ms | |
- | CogVideoX 5B | 5B | INT4 Q8 FP16 | 95 tok/s 61 tok/s 34 tok/s | ~745 ms ~751 ms ~764 ms | |
- | Wan 2.2 TI2V 5B | 5B | INT4 Q8 FP16 | 95 tok/s 61 tok/s 34 tok/s | ~745 ms ~751 ms ~764 ms | |
- | Phi-2 | 2.7B | INT4 Q8 FP16 | 94 tok/s 74 tok/s 48 tok/s | ~425 ms ~428 ms ~435 ms | |
- | MaLLaM-3B | 3B | INT4 Q8 FP16 | 89 tok/s 69 tok/s 45 tok/s | ~467 ms ~470 ms ~478 ms | |
- | SmolLM3 3B | 3B | INT4 Q8 FP16 | 89 tok/s 69 tok/s 45 tok/s | ~467 ms ~470 ms ~478 ms | |
- | Falcon-7B | 7B | INT4 Q8 FP16 | 78 tok/s 47 tok/s 24 tok/s | ~1.0s ~1.0s ~1.1s | |
- | AuraFlow v0.2 | 6.8B | INT4 Q8 FP16 | 76 tok/s 47 tok/s 24 tok/s | ~999 ms ~1.0s ~1.1s | |
- | AuraFlow v0.3 | 6.8B | INT4 Q8 FP16 | 76 tok/s 47 tok/s 24 tok/s | ~999 ms ~1.0s ~1.1s | |
- | Falcon3-7B | 7B | INT4 Q8 FP16 | 74 tok/s 46 tok/s 22 tok/s | ~1.0s ~1.0s ~1.2s | |
- | Hy-MT2 7B | 7B | INT4 Q8 FP16 | 74 tok/s 46 tok/s 22 tok/s | ~1.0s ~1.0s ~1.2s | |
- | Mistral-7B-v0.1 | 7.3B | INT4 Q8 FP16 | 72 tok/s 44 tok/s 22 tok/s | ~1.1s ~1.1s ~1.3s | |
- | Youtu Parsing Omni | 8B | INT4 Q8 | 69 tok/s 41 tok/s | ~1.2s ~1.2s | |
- | Stable Diffusion 3.5 Large | 8.1B | INT4 Q8 | 66 tok/s 40 tok/s | ~1.2s ~1.2s | |
- | Stable Diffusion 3.5 Large Turbo | 8.1B | INT4 Q8 | 66 tok/s 40 tok/s | ~1.2s ~1.2s | |
- | FLUX.2 Klein 9B | 9B | INT4 Q8 | 61 tok/s 37 tok/s | ~1.3s ~1.3s | |
- | LensVLM 9B | 9B | INT4 Q8 | 61 tok/s 37 tok/s | ~1.3s ~1.3s | |
- | MiMo V2.6 Distill Qwen 9B | 9B | INT4 Q8 | 61 tok/s 37 tok/s | ~1.3s ~1.3s | |
- | ChatGLM3-6B-32K | 6B | INT4 Q8 FP16 | 59 tok/s 42 tok/s 24 tok/s | ~891 ms ~898 ms ~991 ms | |
- | Yi-6B | 6B | INT4 Q8 FP16 | 59 tok/s 42 tok/s 24 tok/s | ~891 ms ~898 ms ~991 ms | |
- | Falcon3-10B | 10B | INT4 Q8 | 56 tok/s 33 tok/s | ~1.5s ~1.5s | |
- | Mochi 1 Preview | 10B | INT4 Q8 | 56 tok/s 33 tok/s | ~1.5s ~1.5s | |
- | Falcon2-11B | 11B | INT4 Q8 | 54 tok/s 31 tok/s | ~1.6s ~1.6s | |
- | Hunyuan Lite | 7B | INT4 Q8 | 53 tok/s 37 tok/s | ~1.0s ~1.0s | |
- | Kimi-VL-A3B-Instruct | 16B (3B active) | INT4 | 53 tok/s | ~599 ms | |
- | OLMo 3 7B Base | 7B | INT4 Q8 | 53 tok/s 37 tok/s | ~1.0s ~1.0s | |
- | SEA-LION-7B | 7.1B | INT4 Q8 | 53 tok/s 37 tok/s | ~1.0s ~1.1s | |
- | SEA-LION-7B-Instruct | 7.1B | INT4 Q8 | 53 tok/s 37 tok/s | ~1.0s ~1.1s | |
- | FLUX.1 Canny Dev | 11.9B | INT4 Q8 | 48 tok/s 27 tok/s | ~1.7s ~1.9s | |
- | FLUX.1 Depth Dev | 11.9B | INT4 Q8 | 48 tok/s 27 tok/s | ~1.7s ~1.9s | |
- | FLUX.1 Dev | 11.9B | INT4 Q8 | 48 tok/s 27 tok/s | ~1.7s ~1.9s | |
- | FLUX.1 Fill Dev | 11.9B | INT4 Q8 | 48 tok/s 27 tok/s | ~1.7s ~1.9s | |
- | FLUX.1 Schnell | 11.9B | INT4 Q8 | 48 tok/s 27 tok/s | ~1.7s ~1.9s | |
- | Sahabat-AI-Llama3-8B-Instruct | 8B | INT4 Q8 | 48 tok/s 34 tok/s | ~1.2s ~1.2s | |
- | Typhoon-2-8B | 8B | INT4 Q8 | 48 tok/s 34 tok/s | ~1.2s ~1.2s | |
- | HunyuanVideo 1.5 720P T2V | 13B | INT4 Q8 | 45 tok/s 25 tok/s | ~1.9s ~2.1s | |
- | GLM-4-9B | 9B | INT4 Q8 | 44 tok/s 29 tok/s | ~1.3s ~1.4s | |
- | GLM-4-9B-Chat | 9B | INT4 Q8 | 44 tok/s 29 tok/s | ~1.3s ~1.4s | |
- | GLM-4-9B-Chat-1M | 9B | INT4 Q8 | 44 tok/s 29 tok/s | ~1.3s ~1.4s | |
- | Yi-9B | 9B | INT4 Q8 | 44 tok/s 29 tok/s | ~1.3s ~1.4s | |
- | Sahabat-AI-Gemma2-9B | 9.2B | INT4 Q8 | 43 tok/s 28 tok/s | ~1.3s ~1.5s | |
- | Sahabat-AI-Gemma2-9B-Instruct | 9.2B | INT4 Q8 | 43 tok/s 28 tok/s | ~1.3s ~1.5s | |
- | Wan 2.1 I2V 14B 720P | 14B | INT4 Q8 | 42 tok/s 22 tok/s | ~2.0s ~2.4s | |
- | Wan 2.1 T2V 14B | 14B | INT4 Q8 | 42 tok/s 22 tok/s | ~2.0s ~2.4s | |
- | Wan 2.2 T2V A14B | 14B | INT4 Q8 | 42 tok/s 22 tok/s | ~2.0s ~2.4s | |
- | FLUX.2 Dev | 15B | INT4 Q8 | 39 tok/s 21 tok/s | ~2.2s ~2.6s | |
- | GreenMind-14B-R1 | 14B | INT4 | 31 tok/s | ~2.0s | |
- | Arete 27B | 27B | INT4 | 21 tok/s | ~4.5s | |
- | AREX-2 | 27B | INT4 | 21 tok/s | ~4.5s | |
- | PPLX Decider v1 27B | 27B | INT4 | 21 tok/s | ~4.5s | |
- | PPLX Decider v1.1 27B | 27B | INT4 | 21 tok/s | ~4.5s | |
Inference Capacity
Maximum model parameter class runnable by weight precision.
Context:
4-bit
14B Models
8-bit
8B Models
16-bit
3B Models
Fine-Tuning Capacity
Hardware capacity for local training and adapter fine-tuning.
Context Length:
QLoRA
14B Models
LoRA
8B Models
Full Parameter Training
1B Models
Workload Recommendations
Guidance for optimal precision and operational ceilings on NVIDIA RTX A4000 (16GB).
Inference Sweet Spot
Optimal for 8B models across extended context (32k+). 14B models fit comfortably at Q4 precision with standard context.
Bandwidth & Speed Profile
With 448 GB/s aggregate bandwidth, batch size 1 inference operates in a memory-bandwidth bound regime. Generates approximately 100 tok/s on an 8B Q4 model and 12 tok/s on a 70B Q4 model.
Frequently Asked Questions
©2025 ApX Machine Learning