Model comparison

Olmo-3-7B-Think vs HyperCLOVAX-SEED-Text-Instruct-1.5B

At Q4_K_M, HyperCLOVAX-SEED-Text-Instruct-1.5B is the smaller download — 1,133,974,368 bytes against 4,471,600,192.

From the file· summed bytes, KV per layer

Side by side

Olmo-3-7B-ThinkHyperCLOVAX-SEED-Text-Instruct-1.5B
Parameters7.3B1.6B
Architectureolmo2llama
Layers32
Native context65,536
Mixture of expertsnono
Quantizations published471
Smallest quantization1.69 GiB1.06 GiB
Q4_K_M4.16 GiB1.06 GiB
Licenceapache-2.0other

KV cache by context

the term that decides long-context viability
ContextOlmo-3-7B-ThinkHyperCLOVAX-SEED-Text-Instruct-1.5BRatio
4,0962.00 GiB
8,1922.69 GiB
16,3843.69 GiB
32,7685.69 GiB
65,5369.69 GiB
131,07217.69 GiB