ef9241d1d349ace82d320b0a689d33669833d08f

 1diff --git a/dot_config/llama-server/models.ini b/dot_config/llama-server/models.ini
 2index 5699c9819ae806a032ac0a84d4716c6fac6a4e7c..f8b5b4668f85f178791e35d5cb2be8fe2f05d219 100644
 3--- a/dot_config/llama-server/models.ini
 4+++ b/dot_config/llama-server/models.ini
 5@@ -15,6 +15,28 @@ embedding = true
 6 hf = unsloth/Qwen3.8-27B-GGUF:UD-IQ3_S
 7 alias = qwen3.8-27b-q3,reviewer
 8 n-gpu-layers = auto
 9+n-gpu-layers-draft = 99
10+fit = off
11+flash-attn = on
12+jinja = true
13+ctx-size = 131072
14+temp = 0.6
15+top-p = 0.95
16+top-k = 20
17+min-p = 0.0
18+parallel = 1
19+spec-type = draft-mtp
20+spec-draft-n-max = 2
21+repeat-penalty = 1.0
22+presence-penalty = 1.5
23+reasoning = on
24+chat-template-kwargs = {"reasoning_effort":"medium"}
25+
26+[ukisai/Swift-1.5-Qwen3.8-27B-GGUF:Q3_K_S]
27+hf = ukisai/Swift-1.5-Qwen3.8-27B-GGUF:Q3_K_S
28+alias = swift-1.5-27b-q3
29+n-gpu-layers = auto
30+n-gpu-layers-draft = 99
31 fit = off
32 flash-attn = on
33 jinja = true
34@@ -29,7 +51,7 @@ spec-draft-n-max = 2
35 repeat-penalty = 1.0
36 presence-penalty = 1.5
37 reasoning = on
38-chat-template-kwargs = {"reasoning_effort":"low"}
39+chat-template-kwargs = {"reasoning_effort":"medium"}
40 
41 [unsloth/gemma-4-26B-A4B-it-qat-GGUF:Q4_K_XL]
42 hf = unsloth/gemma-4-26B-A4B-it-qat-GGUF:UD-Q4_K_XL