| NVIDIA RTX PRO 6000 Blackwell Workstation Edition |
96 GB |
600 W |
PCIe CEM5 16-pin |
2 |
GLM-4.5-Air Estimated Q4_K_M,
74.8 GiB ·
18 models fit |
0 |
| NVIDIA A100 80GB PCIe |
80 GB |
300 W |
EPS 8pin (the connector type used for processor power) |
2 |
GLM-4.5-Air Estimated Q4_K_M,
74.8 GiB ·
18 models fit |
0 |
| NVIDIA H100 PCIe |
80 GB |
350 W |
PCIe CEM5 16-pin |
2 |
GLM-4.5-Air Estimated Q4_K_M,
74.8 GiB ·
18 models fit |
0 |
| NVIDIA A40 |
48 GB |
300 W |
EPS 8pin (the connector type used for processor power) |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| NVIDIA L40 |
48 GB |
300 W |
PCIe CEM5 16-pin |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| NVIDIA L40S |
48 GB |
350 W |
PCIe CEM5 16-pin |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| NVIDIA Quadro RTX 8000 |
48 GB |
295 W |
6pin+8pin |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| NVIDIA RTX 6000 Ada Generation |
48 GB |
300 W |
PCIe CEM5 16-pin |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| NVIDIA RTX A6000 |
48 GB |
300 W |
EPS 8pin (the connector type used for processor power) |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
1 |
| NVIDIA A100 40GB PCIe |
40 GB |
250 W |
EPS 8pin (the connector type used for processor power) |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| GeForce RTX 5090 (Reference design) |
32 GB |
575 W |
PCIe CEM5 16-pin |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| NVIDIA RTX 5000 Ada Generation |
32 GB |
250 W |
PCIe CEM5 16-pin |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
1 |
| NVIDIA RTX PRO 4500 Blackwell |
32 GB |
200 W |
PCIe CEM5 16-pin |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| NVIDIA Tesla V100 PCIe 32GB |
32 GB |
250 W |
EPS 8pin (the connector type used for processor power) |
2 |
Qwen3 32B Q4_K_M,
27.5 GiB ·
15 models fit |
0 |
| AMD Radeon RX 7900 XTX (Reference design) |
24 GB |
355 W |
8pin x2 |
2.5 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| GeForce RTX 3090 (Reference design) |
24 GB |
350 W |
8pin x2 |
3 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| GeForce RTX 3090 Ti (Reference design) |
24 GB |
450 W |
PCIe CEM5 16-pin |
3 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| GeForce RTX 4090 (Reference design) |
24 GB |
450 W |
PCIe CEM5 16-pin |
3 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA A10 |
24 GB |
150 W |
8pin |
1 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA A30 |
24 GB |
165 W |
EPS 8pin (the connector type used for processor power) |
2 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA L4 |
24 GB |
72 W |
Not required |
1 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA Quadro M6000 24GB (Reference design) |
24 GB |
250 W |
8pin |
2 |
EXAONE 4.0 32B Estimated Q4_K_M,
21.8 GiB ·
14 models fit |
12 |
| NVIDIA Quadro P6000 (Reference design) |
24 GB |
250 W |
8pin |
2 |
EXAONE 4.0 32B Estimated Q4_K_M,
21.8 GiB ·
14 models fit |
13 |
| NVIDIA Quadro RTX 6000 |
24 GB |
295 W |
6pin+8pin |
2 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA RTX 4500 Ada Generation |
24 GB |
210 W |
Needs checking; the manufacturer does not state it |
2 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA RTX A5000 |
24 GB |
230 W |
8pin |
2 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
13 |
| NVIDIA RTX A5500 |
24 GB |
230 W |
8pin |
2 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
13 |
| NVIDIA RTX PRO 4000 Blackwell |
24 GB |
140 W |
PCIe CEM5 16-pin |
1 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA TITAN RTX (Reference design) |
24 GB |
280 W |
8pin x2 |
2 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA Tesla M40 24GB |
24 GB |
250 W |
EPS 8pin (the connector type used for processor power) |
2 |
EXAONE 4.0 32B Estimated Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| NVIDIA Tesla P40 |
24 GB |
250 W |
EPS 8pin (the connector type used for processor power) |
2 |
EXAONE 4.0 32B Q4_K_M,
21.8 GiB ·
14 models fit |
0 |
| AMD Radeon RX 7900 XT (Reference design) |
20 GB |
315 W |
8pin x2 |
2.5 |
Gemma 3 27B Q4_0 (QAT),
20.0 GiB ·
12 models fit |
0 |
| NVIDIA RTX 4000 Ada Generation |
20 GB |
130 W |
PCIe CEM5 16-pin |
1 |
Gemma 3 27B Q4_0 (QAT),
20.0 GiB ·
12 models fit |
1 |
| NVIDIA RTX 4000 SFF Ada Generation |
20 GB |
70 W |
Not required |
2 |
Gemma 3 27B Q4_0 (QAT),
20.0 GiB ·
12 models fit |
62 |
| NVIDIA RTX A4500 |
20 GB |
200 W |
8pin |
2 |
Gemma 3 27B Q4_0 (QAT),
20.0 GiB ·
12 models fit |
16 |
| AMD Radeon RX 6800 (Reference design) |
16 GB |
250 W |
8pin x2 |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 6800 XT (Reference design) |
16 GB |
300 W |
8pin x2 |
2.5 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 6900 XT (Reference design) |
16 GB |
300 W |
8pin x2 |
2.5 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 6950 XT (Reference design) |
16 GB |
335 W |
8pin x2 |
Not published |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 7600 XT (Reference design) |
16 GB |
190 W |
8pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
19 |
| AMD Radeon RX 7800 XT (Reference design) |
16 GB |
263 W |
8pin x2 |
2.5 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 7900 GRE (Reference design) |
16 GB |
260 W |
8pin x2 |
Not published |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 9060 XT (Reference design) |
16 GB |
160 W |
8pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
19 |
| AMD Radeon RX 9070 (Reference design) |
16 GB |
220 W |
8pin x2 |
Not published |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 9070 XT (Reference design) |
16 GB |
304 W |
8pin x2 |
2.5 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon VII (Reference design) |
16 GB |
295 W |
8pin x2 |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon Vega Frontier Edition (Reference design) |
16 GB |
300 W |
8pin x2 |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| GeForce RTX 4060 Ti (Reference design), 16 GB |
16 GB |
165 W |
8pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
16 |
| GeForce RTX 4070 Ti SUPER (Reference design) |
16 GB |
285 W |
PCIe CEM5 16-pin |
Not published |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| GeForce RTX 4080 (Reference design) |
16 GB |
320 W |
PCIe CEM5 16-pin |
3 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| GeForce RTX 4080 SUPER (Reference design) |
16 GB |
320 W |
PCIe CEM5 16-pin |
3 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| GeForce RTX 5060 Ti (Reference design), 16 GB |
16 GB |
180 W |
8pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
19 |
| GeForce RTX 5070 Ti (Reference design) |
16 GB |
300 W |
PCIe CEM5 16-pin |
Not published |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| GeForce RTX 5080 (Reference design) |
16 GB |
360 W |
PCIe CEM5 16-pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| Intel Arc A770 (Reference design), 16 GB |
16 GB |
225 W |
6pin+8pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| NVIDIA A16 |
16 GB |
250 W |
EPS 8pin (the connector type used for processor power) |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| NVIDIA Quadro P5000 (Reference design) |
16 GB |
180 W |
8pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
15 |
| NVIDIA Quadro RTX 5000 |
16 GB |
265 W |
6pin+8pin |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| NVIDIA RTX 2000 Ada Generation |
16 GB |
70 W |
Not required |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
62 |
| NVIDIA RTX A4000 |
16 GB |
140 W |
6pin |
1 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
20 |
| NVIDIA T4 |
16 GB |
70 W |
Not required |
1 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| NVIDIA Tesla P100 PCIe 16GB |
16 GB |
250 W |
EPS 8pin (the connector type used for processor power) |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| NVIDIA Tesla V100 PCIe 16GB |
16 GB |
250 W |
EPS 8pin (the connector type used for processor power) |
2 |
Phi-4 14B Estimated Q4_K_M,
15.6 GiB ·
10 models fit |
0 |
| AMD Radeon RX 6700 XT (Reference design) |
12 GB |
230 W |
6pin+8pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| AMD Radeon RX 6750 XT (Reference design) |
12 GB |
250 W |
6pin+8pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| AMD Radeon RX 7700 XT (Reference design) |
12 GB |
245 W |
8pin x2 |
2.5 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce GTX TITAN X (Reference design) |
12 GB |
250 W |
6pin+8pin |
2 |
Gemma 3 12B Estimated Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce RTX 2060 12GB (Reference design) |
12 GB |
185 W |
8pin |
Not published |
Gemma 3 12B Estimated Q4_0 (QAT),
10.9 GiB ·
7 models fit |
19 |
| GeForce RTX 3080 (Reference design), 12 GB |
12 GB |
350 W |
8pin x2 |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce RTX 3080 Ti (Reference design) |
12 GB |
350 W |
8pin x2 |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce RTX 4070 (Reference design) |
12 GB |
200 W |
PCIe CEM5 16-pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce RTX 4070 SUPER (Reference design) |
12 GB |
220 W |
PCIe CEM5 16-pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce RTX 4070 Ti (Reference design) |
12 GB |
285 W |
PCIe CEM5 16-pin |
Not published |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce RTX 5070 (Reference design) |
12 GB |
250 W |
PCIe CEM5 16-pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| Intel Arc B580 (Reference design) |
12 GB |
190 W |
8pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
11 |
| NVIDIA Quadro M6000 (Reference design) |
12 GB |
250 W |
8pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
12 |
| NVIDIA RTX A2000 12GB |
12 GB |
70 W |
Not required |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
62 |
| NVIDIA TITAN V (Reference design) |
12 GB |
250 W |
6pin+8pin |
2 |
Gemma 3 12B Estimated Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| NVIDIA TITAN Xp (Reference design) |
12 GB |
250 W |
6pin+8pin |
2 |
Gemma 3 12B Estimated Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| PALIT GeForce RTX 3060 Infinity 2 OC |
12 GB |
170 W |
8pin |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
16 |
| GeForce GTX 1080 Ti (Reference design) |
11 GB |
250 W |
6pin+8pin |
2 |
Gemma 3 12B Estimated Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| GeForce RTX 2080 Ti (Reference design) |
11 GB |
260 W |
8pin x2 |
2 |
Gemma 3 12B Q4_0 (QAT),
10.9 GiB ·
7 models fit |
0 |
| AMD Radeon RX 6700 (Reference design) |
10 GB |
175 W |
8pin |
2 |
Llama 3.1 8B Estimated Q4_K_M,
9.6 GiB ·
5 models fit |
15 |
| GeForce RTX 3080 (Reference design), 10 GB |
10 GB |
320 W |
8pin x2 |
2 |
Llama 3.1 8B Estimated Q4_K_M,
9.6 GiB ·
5 models fit |
0 |
| Intel Arc B570 (Reference design) |
10 GB |
150 W |
8pin |
2 |
Llama 3.1 8B Estimated Q4_K_M,
9.6 GiB ·
5 models fit |
19 |
| AMD Radeon R9 390 (Reference design) |
8 GB |
275 W |
6pin+8pin |
2 |
Qwen3 4B Estimated Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| AMD Radeon RX 480 (Reference design) |
8 GB |
150 W |
6pin |
2 |
Qwen3 4B Estimated Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 5500 XT (Reference design), 8 GB |
8 GB |
130 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 5700 (Reference design) |
8 GB |
180 W |
6pin+8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| AMD Radeon RX 5700 XT (Reference design) |
8 GB |
225 W |
6pin+8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| AMD Radeon RX 580 (Reference design) |
8 GB |
185 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 590 (Reference design) |
8 GB |
175 W |
8pin |
2 |
Qwen3 4B Estimated Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 6600 (Reference design) |
8 GB |
132 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 6600 XT (Reference design) |
8 GB |
160 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 6650 XT (Reference design) |
8 GB |
180 W |
8pin |
2 |
Qwen3 4B Estimated Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 7600 (Reference design) |
8 GB |
165 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX 9050 (Reference design) |
8 GB |
92 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| AMD Radeon RX Vega 56 (Reference design) |
8 GB |
210 W |
8pin x2 |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| AMD Radeon RX Vega 64 (Reference design) |
8 GB |
295 W |
8pin x2 |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| GeForce GTX 1070 (Reference design) |
8 GB |
150 W |
8pin |
2 |
Qwen3 4B Estimated Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce GTX 1070 Ti (Reference design) |
8 GB |
180 W |
8pin |
2 |
Qwen3 4B Estimated Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce GTX 1080 (Reference design) |
8 GB |
180 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce RTX 2060 SUPER (Reference design) |
8 GB |
175 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
16 |
| GeForce RTX 2070 (Reference design) |
8 GB |
185 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
16 |
| GeForce RTX 2070 SUPER (Reference design) |
8 GB |
215 W |
6pin+8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| GeForce RTX 2080 (Reference design) |
8 GB |
225 W |
6pin+8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| GeForce RTX 2080 SUPER (Reference design) |
8 GB |
250 W |
6pin+8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| GeForce RTX 3050 8GB (Reference design) |
8 GB |
130 W |
8pin |
Not published |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce RTX 3060 (Reference design) |
8 GB |
170 W |
8pin |
Not published |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce RTX 3060 Ti (Reference design) |
8 GB |
200 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
16 |
| GeForce RTX 3070 (Reference design) |
8 GB |
220 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
13 |
| GeForce RTX 3070 Ti (Reference design) |
8 GB |
290 W |
8pin x2 |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| GeForce RTX 4060 (Reference design) |
8 GB |
115 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
20 |
| GeForce RTX 4060 OC Low Profile 8G (GV-N4060OC-8GL) |
8 GB |
115 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
17 |
| GeForce RTX 4060 Ti (Reference design), 8 GB |
8 GB |
160 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
16 |
| GeForce RTX 5050 (Reference design) |
8 GB |
130 W |
8pin |
Not published |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce RTX 5060 (Reference design) |
8 GB |
145 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce RTX 5060 Ti (Reference design), 8 GB |
8 GB |
180 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| GeForce RTX 5060 Ti 8G VENTUS 2X PLUS |
8 GB |
180 W |
8pin |
2.5 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
15 |
| Intel Arc A580 (Reference design) |
8 GB |
185 W |
8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| Intel Arc A750 (Reference design) |
8 GB |
225 W |
6pin+8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| Intel Arc A770 (Reference design), 8 GB |
8 GB |
225 W |
6pin+8pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| NVIDIA Quadro M4000 (Reference design) |
8 GB |
120 W |
6pin |
1 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| NVIDIA Quadro M5000 (Reference design) |
8 GB |
150 W |
6pin |
2 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
18 |
| NVIDIA Quadro P4000 (Reference design) |
8 GB |
105 W |
6pin |
1 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
19 |
| NVIDIA Quadro RTX 4000 |
8 GB |
160 W |
8pin |
1 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
16 |
| NVIDIA RTX A1000 |
8 GB |
50 W |
Not required |
1 |
Qwen3 4B Q4_K_M,
7.9 GiB ·
4 models fit |
65 |
| NVIDIA Tesla P4 |
8 GB |
75 W |
Not required |
1 |
Qwen3 4B Estimated Q4_K_M,
7.9 GiB ·
4 models fit |
0 |
| AMD Radeon RX 5600 XT (Reference design) |
6 GB |
150 W |
8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
19 |
| GeForce GTX 1060 (Reference design) |
6 GB |
120 W |
6pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
22 |
| GeForce GTX 1660 (Reference design) |
6 GB |
120 W |
8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
19 |
| GeForce GTX 1660 SUPER (Reference design) |
6 GB |
125 W |
8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
19 |
| GeForce GTX 1660 Ti (Reference design) |
6 GB |
120 W |
8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
19 |
| GeForce GTX 980 Ti (Reference design) |
6 GB |
250 W |
6pin+8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
0 |
| GeForce GTX TITAN (Reference design) |
6 GB |
250 W |
6pin+8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
0 |
| GeForce GTX TITAN Black (Reference design) |
6 GB |
250 W |
6pin+8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
0 |
| GeForce RTX 2060 (Reference design) |
6 GB |
160 W |
8pin |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
19 |
| GeForce RTX 3050 6GB (Reference design) |
6 GB |
70 W |
Not required |
Not published |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
50 |
| GeForce RTX 3050 OC Low Profile 6G (GV-N3050OC-6GL) |
6 GB |
70 W |
Not required |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
54 |
| Intel Arc A380 (Reference design) |
6 GB |
75 W |
Not required |
1 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
82 |
| NVIDIA RTX A2000 6GB |
6 GB |
70 W |
Not required |
2 |
Qwen3 1.7B Estimated Q4_K_M,
5.6 GiB ·
3 models fit |
62 |
| NVIDIA Quadro P2000 (Reference design) |
5 GB |
75 W |
Not required |
1 |
Qwen3 0.6B Estimated Q4_K_M,
4.9 GiB ·
2 models fit |
49 |
| NVIDIA Quadro P2200 (Reference design) |
5 GB |
75 W |
Not required |
1 |
Qwen3 0.6B Estimated Q4_K_M,
4.9 GiB ·
2 models fit |
49 |
| AMD Radeon R9 290 (Reference design) |
4 GB |
275 W |
6pin+8pin |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon R9 290X (Reference design) |
4 GB |
290 W |
6pin+8pin |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon R9 Fury (Reference design) |
4 GB |
275 W |
8pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon R9 Fury X (Reference design) |
4 GB |
275 W |
8pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon R9 Nano (Reference design) |
4 GB |
175 W |
8pin |
2 |
No model in this database fits at the working context length |
19 |
| AMD Radeon RX 470 (Reference design) |
4 GB |
120 W |
6pin |
2 |
No model in this database fits at the working context length |
19 |
| AMD Radeon RX 550 (Reference design) |
4 GB |
50 W |
Not required |
2 |
No model in this database fits at the working context length |
80 |
| AMD Radeon RX 5500 XT (Reference design), 4 GB |
4 GB |
130 W |
8pin |
2 |
No model in this database fits at the working context length |
19 |
| AMD Radeon RX 560 (Reference design) |
4 GB |
75 W |
Not required |
2 |
No model in this database fits at the working context length |
50 |
| AMD Radeon RX 570 (Reference design) |
4 GB |
150 W |
6pin |
2 |
No model in this database fits at the working context length |
22 |
| AMD Radeon RX 6500 XT (Reference design) |
4 GB |
107 W |
6pin |
2 |
No model in this database fits at the working context length |
22 |
| GeForce GTX 1050 Ti (Reference design) |
4 GB |
75 W |
Not required |
2 |
No model in this database fits at the working context length |
50 |
| GeForce GTX 1630 (Reference design) |
4 GB |
75 W |
Not required |
1 |
No model in this database fits at the working context length |
82 |
| GeForce GTX 1650 (Reference design) |
4 GB |
75 W |
Needs checking; varies by board |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 1650 SUPER (Reference design) |
4 GB |
100 W |
6pin |
2 |
No model in this database fits at the working context length |
19 |
| GeForce GTX 970 (Reference design) |
4 GB |
148 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 980 (Reference design) |
4 GB |
165 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| Intel Arc A310 (Reference design) |
4 GB |
75 W |
Not required |
1 |
No model in this database fits at the working context length |
82 |
| NVIDIA Quadro K1200 (Reference design) |
4 GB |
45 W |
Not required |
1 |
No model in this database fits at the working context length |
65 |
| NVIDIA Quadro K2200 (Reference design) |
4 GB |
68 W |
Not required |
1 |
No model in this database fits at the working context length |
49 |
| NVIDIA Quadro K4200 (Reference design) |
4 GB |
108 W |
6pin |
1 |
No model in this database fits at the working context length |
19 |
| NVIDIA Quadro M2000 (Reference design) |
4 GB |
75 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| NVIDIA Quadro P1000 (Reference design) |
4 GB |
47 W |
Not required |
1 |
No model in this database fits at the working context length |
82 |
| NVIDIA RTX A400 |
4 GB |
50 W |
Not required |
1 |
No model in this database fits at the working context length |
65 |
| NVIDIA T1000 (Reference design) |
4 GB |
50 W |
Not required |
1 |
No model in this database fits at the working context length |
65 |
| NVIDIA T600 |
4 GB |
40 W |
Not required |
1 |
No model in this database fits at the working context length |
65 |
| RD-RX6400-E4GB/LP |
4 GB |
53 W |
Not required |
1 |
No model in this database fits at the working context length |
77 |
| AMD Radeon R9 280 (Reference design) |
3 GB |
250 W |
6pin+8pin |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon R9 280X (Reference design) |
3 GB |
250 W |
6pin+8pin |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 780 (Reference design) |
3 GB |
250 W |
6pin+8pin |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 780 Ti (Reference design) |
3 GB |
250 W |
6pin+8pin |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon R7 240 (Reference design) |
2 GB |
50 W |
Not required |
2 |
No model in this database fits at the working context length |
80 |
| AMD Radeon R7 250 (Reference design) |
2 GB |
75 W |
Not required |
2 |
No model in this database fits at the working context length |
50 |
| AMD Radeon R7 260X (Reference design) |
2 GB |
115 W |
6pin |
2 |
No model in this database fits at the working context length |
22 |
| AMD Radeon R9 270 (Reference design) |
2 GB |
150 W |
6pin |
2 |
No model in this database fits at the working context length |
22 |
| AMD Radeon R9 270X (Reference design) |
2 GB |
180 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon R9 380 (Reference design) |
2 GB |
190 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| AMD Radeon RX 460 (Reference design) |
2 GB |
75 W |
Not required |
2 |
No model in this database fits at the working context length |
50 |
| GeForce GT 1010 (Reference design) |
2 GB |
30 W |
Not required |
Not published |
No model in this database fits at the working context length |
50 |
| GeForce GT 1030 (Reference design) |
2 GB |
30 W |
Not required |
1 |
No model in this database fits at the working context length |
81 |
| GeForce GT 630 (Reference design) |
2 GB |
25 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| GeForce GT 640 (Reference design) |
2 GB |
65 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| GeForce GT 710 (Reference design) |
2 GB |
19 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| GeForce GT 730 (Reference design) |
2 GB |
38 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| GeForce GTX 1050 (Reference design) |
2 GB |
75 W |
Not required |
2 |
No model in this database fits at the working context length |
50 |
| GeForce GTX 660 (Reference design) |
2 GB |
140 W |
6pin |
2 |
No model in this database fits at the working context length |
19 |
| GeForce GTX 660 Ti (Reference design) |
2 GB |
150 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 670 (Reference design) |
2 GB |
170 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 680 (Reference design) |
2 GB |
195 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 750 Ti (Reference design) |
2 GB |
60 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| GeForce GTX 760 (Reference design) |
2 GB |
170 W |
6pin x2 |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 770 (Reference design) |
2 GB |
230 W |
6pin+8pin |
2 |
No model in this database fits at the working context length |
0 |
| GeForce GTX 950 (Reference design) |
2 GB |
90 W |
6pin |
2 |
No model in this database fits at the working context length |
22 |
| GeForce GTX 960 (Reference design) |
2 GB |
120 W |
6pin |
2 |
No model in this database fits at the working context length |
19 |
| NVIDIA Quadro K620 (Reference design) |
2 GB |
45 W |
Not required |
1 |
No model in this database fits at the working context length |
65 |
| NVIDIA Quadro P400 (Reference design) |
2 GB |
30 W |
Not required |
1 |
No model in this database fits at the working context length |
82 |
| NVIDIA Quadro P600 (Reference design) |
2 GB |
40 W |
Not required |
1 |
No model in this database fits at the working context length |
82 |
| NVIDIA Quadro P620 (Reference design) |
2 GB |
40 W |
Not required |
1 |
No model in this database fits at the working context length |
82 |
| NVIDIA T400, 2 GB |
2 GB |
30 W |
Not required |
1 |
No model in this database fits at the working context length |
65 |
| GeForce GT 610 (Reference design) |
1 GB |
29 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| GeForce GT 620 (Reference design) |
1 GB |
49 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| GeForce GTX 650 (Reference design) |
1 GB |
64 W |
6pin |
2 |
No model in this database fits at the working context length |
22 |
| GeForce GTX 650 Ti (Reference design) |
1 GB |
110 W |
6pin |
2 |
No model in this database fits at the working context length |
22 |
| GeForce GTX 750 (Reference design) |
1 GB |
55 W |
Not required |
1 |
No model in this database fits at the working context length |
50 |
| NVIDIA Quadro K600 (Reference design) |
1 GB |
41 W |
Not required |
1 |
No model in this database fits at the working context length |
65 |
