Model comparison
GLM-4.6V-Flash vs Qwen3-8B
At Q4_K_M, Qwen3-8B is the smaller download — 5,027,783,488 bytes against 6,166,577,888. At long context the gap widens: GLM-4.6V-Flash's KV cache at 32K is 3.6× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| GLM-4.6V-Flash | Qwen3-8B | |
|---|---|---|
| Parameters | 10.3B | 8.2B |
| Architecture | glm4 | qwen3 |
| Layers | 40 | 36 |
| Native context | 131,072 | 40,960 |
| Mixture of experts | no | no |
| Quantizations published | 47 | 51 |
| Smallest quantization | 2.98 GiB | 2.12 GiB |
| Q4_K_M | 5.74 GiB | 4.68 GiB |
| Licence | mit | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | GLM-4.6V-Flash | Qwen3-8B | Ratio |
|---|---|---|---|
| 4,096 | 0.16 GiB | 0.56 GiB | 3.60× |
| 8,192 | 0.31 GiB | 1.13 GiB | 3.60× |
| 16,384 | 0.63 GiB | 2.25 GiB | 3.60× |
| 32,768 | 1.25 GiB | 4.50 GiB | 3.60× |
| 65,536 | 2.50 GiB | 9.00 GiB | 3.60× |
| 131,072 | 5.00 GiB | 18.00 GiB | 3.60× |