frank-large: pin Qwen3.8-Flash-Next sampler settings + preserve_thinking

halogen-flash-server has no server-side sampler defaults (temperature absent
or 0 = greedy), so the endpoint decoded greedy for OMP, whose global
sampling settings default to -1 (not sent).

compat.extraBody merges these top-level into every request body after the
compat policy, so they win:
  temperature=1.0 top_p=0.95 top_k=20 min_p=0.0
  presence_penalty=0.0 repetition_penalty=1.0 preserve_thinking=true

Verified against the live dankclaw endpoint: preserve_thinking is honoured
(350 vs 75 prompt tokens for a replayed reasoning turn), and the exact
body shape returns 200.

agent/models.yml

@@ -24,13 +24,26 @@ providers:
     models:
       - id: frank-large
         name: "frank-large (dankclaw-gpu)"
-        contextWindow: 200000
+        contextWindow: 262144
-        maxTokens: 200000
+        maxTokens: 262144
         reasoning: true
-        input: [text]
+        input: [text, image]
         compat:
           reasoningEffortMap:
             max: xhigh
+          # Qwen3.8-Flash-Next recommended sampling; halogen has no server-side
+          # sampler defaults, so without this the endpoint decodes greedy.
+          # extraBody is merged last (after compat policy) and wins.
+          extraBody:
+            temperature: 1.0
+            top_p: 0.95
+            top_k: 20
+            min_p: 0.0
+            presence_penalty: 0.0
+            repetition_penalty: 1.0
+            # Keep prior thinking blocks in the rendered prompt across turns;
+            # halogen honours it top-level (verified: 275-token prompt delta vs false).
+            preserve_thinking: true
 
   "flm (dankclaw-npu)":
     baseUrl: http://192.168.0.143:7771/v1