Model comparison
Le-Chaton-Slim-23B vs Qwen3-Coder-30B-A3B-Instruct
At Q4_K_M, Le-Chaton-Slim-23B is the smaller download — 14,183,211,648 bytes against 18,556,689,568.
From the file· summed bytes, KV per layer
Side by side
| Le-Chaton-Slim-23B | Qwen3-Coder-30B-A3B-Instruct | |
|---|---|---|
| Parameters | 23.3B | 30.5B |
| Architecture | llama | qwen3moe |
| Layers | 26 | 48 |
| Native context | 262,144 | 262,144 |
| Mixture of experts | yes, 10 experts | yes, 128 experts |
| Quantizations published | 34 | 46 |
| Smallest quantization | 4.59 GiB | 7.46 GiB |
| Q4_K_M | 13.21 GiB | 17.28 GiB |
| Licence | apache-2.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | Le-Chaton-Slim-23B | Qwen3-Coder-30B-A3B-Instruct | Ratio |
|---|---|---|---|
| 4,096 | 0.41 GiB | 0.38 GiB | 1.08× |
| 8,192 | 0.81 GiB | 0.75 GiB | 1.08× |
| 16,384 | 1.63 GiB | 1.50 GiB | 1.08× |
| 32,768 | 3.25 GiB | 3.00 GiB | 1.08× |
| 65,536 | 6.50 GiB | 6.00 GiB | 1.08× |
| 131,072 | 13.00 GiB | 12.00 GiB | 1.08× |