internlm

/

Intern-S1-mini

like 102

Follow

Intern Large Models 839

Not tried vision yet.
As a .gguf quant a Thinkpad T495 (ryzen 3500u, Vega 8) gets:

| Qwen3-8B-Intern-S1-mini.i1-Q4_K_M.gguf                          | qwen3 8B Q4_K - Medium         |   4.68 GiB |     8.20 B | Vulkan     |   0 |       5 |           tg128 |          3.58 ± 0.01 |
| Qwen3-8B-Intern-S1-mini.i1-Q4_K_M.gguf                          | qwen3 8B Q4_K - Medium         |   4.68 GiB |     8.20 B | Vulkan     |  30 |       5 |           pp512 |         12.06 ± 0.04 |
| Qwen3-8B-Intern-S1-mini.i1-Q4_K_M.gguf                          | qwen3 8B Q4_K - Medium         |   4.68 GiB |     8.20 B | Vulkan     |  30 |       5 |           tg128 |          4.22 ± 0.05 |
| Qwen3-8B-Intern-S1-mini.i1-Q4_K_M.gguf                          | qwen3 8B Q4_K - Medium         |   4.68 GiB |     8.20 B | Vulkan     |  99 |       5 |           pp512 |         26.54 ± 0.14 |
| Qwen3-8B-Intern-S1-mini.i1-Q4_K_M.gguf                          | qwen3 8B Q4_K - Medium         |   4.68 GiB |     8.20 B | Vulkan     |  99 |       5 |           tg128 |          4.47 ± 0.01 |

And leveraging Qwen3's advanced reasoning patterns, this is one of the few small models to get the 'chess puzzle' right:
Let's warm up with a logic puzzle: Alice, Bertha and Cindy are in a room cut off from the outside world. The only things in the room are a book and a chess set. Alice is reading the book. Bertha is playing a game of chess. What is Cindy doing?