Is OvisOCR2 or Qwen3-VL-2B-Instruct smaller?

OvisOCR2 is smaller at Q4_K_M: 0.49 GiB against 1.03 GiB.

Which uses less memory at long context?

OvisOCR2 — its KV cache at 32K context is 9.3x smaller, because of how its attention layers cache.

Model comparison

OvisOCR2 vs Qwen3-VL-2B-Instruct

At Q4_K_M, OvisOCR2 is the smaller download — 529,297,280 bytes against 1,107,409,952. At long context the gap widens: OvisOCR2's KV cache at 32K is 9.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	OvisOCR2	Qwen3-VL-2B-Instruct
Parameters	853M	2.1B
Architecture	qwen35	qwen3vl
Layers	24	28
Native context	262,144	262,144
Mixture of experts	no	no
Quantizations published	43	46
Smallest quantization	0.35 GiB	0.50 GiB
Q4_K_M	0.49 GiB	1.03 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	OvisOCR2	Qwen3-VL-2B-Instruct	Ratio
4,096	0.05 GiB	0.44 GiB	9.33×
8,192	0.09 GiB	0.88 GiB	9.33×
16,384	0.19 GiB	1.75 GiB	9.33×
32,768	0.38 GiB	3.50 GiB	9.33×
65,536	0.75 GiB	7.00 GiB	9.33×
131,072	1.50 GiB	14.00 GiB	9.33×