style: Cleaned up comments and organization of quirks.yaml file

quirks.yaml

@@ -4,9 +4,9 @@
 # model data: oauth blocks, request patches, wire-name aliases, behavioral
 # flags, and models that no public catalog provides.
 #
-# `models:` entries are self-complete — `rules` apply ONLY to API-sourced
+# `models:` entries are self-complete. `rules` apply ONLY to API-sourced
 # models, never to hand-owned `models:` entries. Legacy verbatim-preserved
-# entries are grouped under a `# legacy — pruning candidates` comment.
+# entries are grouped under a `# legacy -- pruning candidates` comment.
 
 - provider: openai
   exclude:
@@ -78,7 +78,7 @@
       max_tokens_per_chunk: 8191
       default_chunk_size: 2000
       max_batch_size: 100
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: o4-mini-high
       max_output_tokens: 100000
       real_name: o4-mini
@@ -179,7 +179,7 @@
       max_tokens_per_chunk: 2048
       default_chunk_size: 1500
       max_batch_size: 100
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: gemini-3-pro-preview
       max_input_tokens: 1048576
       supports_vision: true
@@ -296,7 +296,7 @@
             type: enabled
             budget_tokens: 16000
   models:
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: claude-opus-4-1-20250805
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -324,7 +324,7 @@
       input_price: 0.1
       max_tokens_per_chunk: 8092
       default_chunk_size: 2000
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: magistral-small-latest
       max_input_tokens: 131072
       input_price: 0.5
@@ -451,7 +451,7 @@
 
 - provider: perplexity
   models:
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: sonar-reasoning
       max_input_tokens: 128000
       input_price: 1
@@ -461,7 +461,7 @@
   exclude:
     - "whisper-*"
   models:
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: meta-llama/llama-4-maverick-17b-128e-instruct
       max_input_tokens: 131072
       input_price: 0
@@ -600,7 +600,7 @@
       max_tokens_per_chunk: 2048
       default_chunk_size: 1500
       max_batch_size: 5
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: gemini-3-pro-preview
       max_input_tokens: 1048576
       supports_vision: true
@@ -813,7 +813,7 @@
       max_tokens_per_chunk: 512
       default_chunk_size: 1000
       max_batch_size: 96
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: us.meta.llama3-3-70b-instruct-v1:0
       max_input_tokens: 131072
       max_output_tokens: 8192
@@ -855,7 +855,7 @@
       max_tokens_per_chunk: 512
       default_chunk_size: 1000
       max_batch_size: 100
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: '@cf/google/gemma-3-12b-it'
       max_input_tokens: 131072
       max_output_tokens: 2048
@@ -939,7 +939,7 @@
       max_tokens_per_chunk: 8192
       default_chunk_size: 2000
       max_batch_size: 10
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: qwen3-vl-flash
       max_input_tokens: 262144
       supports_vision: true
@@ -1000,7 +1000,7 @@
 
 - provider: deepseek
   models:
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: deepseek-chat
       max_input_tokens: 64000
       max_output_tokens: 8192
@@ -1040,7 +1040,7 @@
       type: reranker
       max_input_tokens: 4096
       input_price: 0.112
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: glm-4.6v-flash
       max_input_tokens: 65536
       input_price: 0
@@ -1049,7 +1049,7 @@
 
 - provider: minimax
   models:
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: minimax-m3
       max_input_tokens: 1000000
       input_price: 4.2
@@ -1166,7 +1166,7 @@
         body:
           include_reasoning: true
   models:
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: google/gemini-3-pro-preview
       max_input_tokens: 1048576
       supports_vision: true
@@ -1352,7 +1352,7 @@
       max_tokens_per_chunk: 512
       default_chunk_size: 1000
       max_batch_size: 100
-    # legacy — pruning candidates
+    # legacy -- pruning candidates
     - name: Qwen/Qwen3-Max-Thinking
       max_input_tokens: 262144
       input_price: 1.2