Model comparison

Seed-OSS-36B-Base vs Qwen3-30B-A3B-Thinking-2507

At Q4_K_M, Qwen3-30B-A3B-Thinking-2507 is the smaller download — 18,556,685,824 bytes against 21,762,145,216. At long context the gap widens: Qwen3-30B-A3B-Thinking-2507's KV cache at 32K is 2.7× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Seed-OSS-36B-BaseQwen3-30B-A3B-Thinking-2507
Parameters36.2B30.5B
Architectureseed_ossqwen3moe
Layers6448
Native context524,288262,144
Mixture of expertsnoyes, 128 experts
Quantizations published551
Smallest quantization16.41 GiB7.05 GiB
Q4_K_M20.27 GiB17.28 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextSeed-OSS-36B-BaseQwen3-30B-A3B-Thinking-2507Ratio
4,0961.00 GiB0.38 GiB2.67×
8,1922.00 GiB0.75 GiB2.67×
16,3844.00 GiB1.50 GiB2.67×
32,7688.00 GiB3.00 GiB2.67×
65,53616.00 GiB6.00 GiB2.67×
131,07232.00 GiB12.00 GiB2.67×
Seed-OSS-36B-Base vs Qwen3-30B-A3B-Thinking-2507 — size, memory and hardware fit — ossmodeldb