Model comparison

Nemotron-3-Embed-8B-BF16 vs Qwen3-Embedding-8B

At Q4_K_M, Qwen3-Embedding-8B is the smaller download — 4,676,804,928 bytes against 4,896,389,984.

From the file· summed bytes, KV per layer

Side by side

Nemotron-3-Embed-8B-BF16Qwen3-Embedding-8B
Parameters8.0B7.6B
Architecturemistral3qwen3
Layers3436
Native context262,14440,960
Mixture of expertsnono
Quantizations published3617
Smallest quantization1.81 GiB2.87 GiB
Q4_K_M4.56 GiB4.36 GiB
Licenceopenmdw-1.1apache-2.0

KV cache by context

the term that decides long-context viability
ContextNemotron-3-Embed-8B-BF16Qwen3-Embedding-8BRatio
4,0960.53 GiB0.56 GiB1.06×
8,1921.06 GiB1.13 GiB1.06×
16,3842.13 GiB2.25 GiB1.06×
32,7684.25 GiB4.50 GiB1.06×
65,5368.50 GiB9.00 GiB1.06×
131,07217.00 GiB18.00 GiB1.06×