Check for model max context

src/server/router.ts

@@ -73,16 +73,20 @@ router.post('/stream', async (req, res) => {
   // const usageAllowed = await checkUsage(user, service)
   // if (!usageAllowed) return res.status(403).send('Usage limit reached')
 
+  const encoding = getEncoding(options.model)
+  let tokenCount = calculateUsage(options, encoding)
+
+  // gpt-3.5-turbo has maximum context of 4096 tokens
+  if (tokenCount > 4000)
+    return res.status(403).send('Model maximum context reached')
+
   options.user = hashData(user.id)
   const stream = await completionStream(options)
 
   if (isError(stream)) return res.status(424).send(stream)
 
   res.setHeader('content-type', 'text/plain')
 
-  const encoding = getEncoding(options.model)
-
-  let tokenCount = calculateUsage(options, encoding)
   // https://github.com/openai/openai-node/issues/18#issuecomment-1493132878
   stream.on('data', (chunk: Buffer) => {
     // Messages in the event stream are separated by a pair of newline characters.