A/B: show reasoning again (remove no_think prefill); raise A/B max_tokens 96->256 so thinking completes

app.py

@@ -127,9 +127,9 @@ async def chat_ab(req: Request):
 
     async def gen():
         try:
-            async for chunk in run_one(BONSAI_BACKEND, "bonsai", messages, mt, no_think=True):
+            async for chunk in run_one(BONSAI_BACKEND, "bonsai", messages, mt):
                 yield chunk
-            async for chunk in run_one(BACKEND, "pocket", messages, mt, no_think=True):
+            async for chunk in run_one(BACKEND, "pocket", messages, mt):
                 yield chunk
             yield _sse({"ab_done": True})
         except Exception as e:

index.html

@@ -248,7 +248,7 @@ async function race(text){
   const acc={bonsai:'',pocket:''}, cnt={bonsai:0,pocket:0}, t0={bonsai:0,pocket:0}, spd={bonsai:0,pocket:0};
   try{
     const r=await fetch('/api/chat_ab',{method:'POST',headers:{'Content-Type':'application/json'},
-      body:JSON.stringify({messages:[{role:'user',content:text}],max_tokens:96})});
+      body:JSON.stringify({messages:[{role:'user',content:text}],max_tokens:256})});
     const rd=r.body.getReader(); const dec=new TextDecoder(); let buf='';
     while(true){ const {done,value}=await rd.read(); if(done) break;
       buf+=dec.decode(value,{stream:true}); const parts=buf.split('\n\n'); buf=parts.pop();