Fara1.5-27B on DGX Spark — BF16, vLLM vision-language inference benchmark

Open this configuration in the VLM Inference Benchmark Explorer

Model	Fara1.5-27B
Parameters	27B
Device	DGX Spark
Quantization	BF16
Image size	720p
Images per camera	1
Max Cameras (response ≤ 3 s)	8
Engine	vLLM

Cameras	Response (s)	TPS (tok/s)	Target
1	0.85	4.5	≤3 s
2	1.23	4.3	≤3 s
4	1.68	4.2	≤3 s
8	2.41	3.8	≤3 s
16	5.87	3.4	>3 s
32	8.67	2.6	>3 s
40	10.6	2.3	>3 s
44	—	—	Failed

Images per request	Response (s)	TPS (tok/s)	Target
1	0.85	4.5	≤3 s
2	1.58	4.5	≤3 s
4	3.00	4.5	≤3 s
8	5.94	4.5	>3 s
16	12.3	4.4	>3 s
32	25.5	4.3	>3 s
64	54.4	4.2	>3 s
128	127.4	3.9	>3 s
256	—	—	Failed

Fara1.5-27B (27B parameters), served in BF16 format with vLLM on DGX Spark, starts answering a single camera that sends one 720p image after 0.85 s, then writes 4.5 tok/s.

With 128 images per camera at 720p, the answer starts after 127.44 s. With one 2K image, after 3.4 s.

At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with 8 cameras at once.