Model comparison
GLM-4.7-Flash-REAP-23B-A3B vs gemma-4-E4B-it
At Q4_K_M, gemma-4-E4B-it is the smaller download — 4,977,171,584 bytes against 14,113,838,816. At long context the gap widens: gemma-4-E4B-it's KV cache at 32K is 3.3× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| GLM-4.7-Flash-REAP-23B-A3B | gemma-4-E4B-it | |
|---|---|---|
| Parameters | 23.0B | 8.0B |
| Architecture | deepseek2 | gemma4 |
| Layers | 47 | 42 |
| Native context | 202,752 | 131,072 |
| Mixture of experts | yes, 48 experts | no |
| Quantizations published | 21 | 36 |
| Smallest quantization | 6.09 GiB | 3.30 GiB |
| Q4_K_M | 13.14 GiB | 4.64 GiB |
| Licence | mit | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | GLM-4.7-Flash-REAP-23B-A3B | gemma-4-E4B-it | Ratio |
|---|---|---|---|
| 4,096 | 0.21 GiB | 0.12 GiB | 1.68× |
| 8,192 | 0.41 GiB | 0.18 GiB | 2.32× |
| 16,384 | 0.83 GiB | 0.29 GiB | 2.88× |
| 32,768 | 1.65 GiB | 0.51 GiB | 3.27× |
| 65,536 | 3.30 GiB | 0.94 GiB | 3.50× |
| 131,072 | 6.61 GiB | 1.82 GiB | 3.63× |