o1 200k context

src/proxy/middleware/request/preprocessors/validate-context-size.ts

@@ -68,9 +68,11 @@ export const validateContextSize: RequestPreprocessor = async (req) => {
     modelMax = 131072;
   } else if (model.match(/^gpt-4(-\d{4})?-vision(-preview)?$/)) {
     modelMax = 131072;
+  } else if (model.match(/^o1(-\d{4}-\d{2}-\d{2})?$/)) {
+    modelMax = 200000;
   } else if (model.match(/^o1-mini(-\d{4}-\d{2}-\d{2})?$/)) {
     modelMax = 128000;
-  } else if (model.match(/^o1(-preview)?(-\d{4}-\d{2}-\d{2})?$/)) {
+  } else if (model.match(/^o1-preview(-\d{4}-\d{2}-\d{2})?$/)) {
     modelMax = 128000;
   } else if (model.match(/gpt-3.5-turbo/)) {
     modelMax = 16384;

src/shared/api-schemas/anthropic.ts

@@ -82,7 +82,7 @@ function openAIMessagesToClaudeTextPrompt(messages: OpenAIChatMessage[]) {
         let role: string = m.role;
         if (role === "assistant") {
           role = "Assistant";
-        } else if (["system", "developer"].includes(role)) {
+        } else if (role === "system") {
           role = "System";
         } else if (role === "user") {
           role = "Human";
@@ -365,7 +365,7 @@ function openAIMessagesToClaudeChatPrompt(messages: OpenAIChatMessage[]): {
     // Here we will lose the original name if it was a system message, but that
     // is generally okay because the system message is usually a prompt and not
     // a character in the chat.
-    const name = ["system", "developer"].includes(msg.role) ? "System" : msg.name?.trim();
+    const name = msg.role === "system" ? "System" : msg.name?.trim();
     const content = convertOpenAIContent(msg.content);
 
     // Prepend the display name to the first text content in the current message
@@ -395,8 +395,8 @@ function openAIMessagesToClaudeChatPrompt(messages: OpenAIChatMessage[]): {
 
 function isSystemOpenAIRole(
   role: OpenAIChatMessage["role"]
-): role is "developer" | "system" | "function" | "tool" {
-  return ["developer","system", "function", "tool"].includes(role);
+): role is "system" | "function" | "tool" {
+  return ["system", "function", "tool"].includes(role);
 }
 
 function getFirstTextContent(content: OpenAIChatMessage["content"]) {

src/shared/api-schemas/openai.ts

@@ -21,7 +21,7 @@ export const OpenAIV1ChatCompletionSchema = z
     model: z.string().max(100),
     messages: z.array(
       z.object({
-        role: z.enum(["developer", "system", "user", "assistant", "tool", "function"]),
+        role: z.enum(["system", "user", "assistant", "tool", "function"]),
         content: z.union([z.string(), OpenAIV1ChatContentArraySchema]),
         name: z.string().optional(),
         tool_calls: z.array(z.any()).optional(),
@@ -78,7 +78,6 @@ export const OpenAIV1ChatCompletionSchema = z
     tool_choice: z.any().optional(),
     function_choice: z.any().optional(),
     response_format: z.any(),
-    reasoning_effort: z.enum(["low", "medium", "high"]).optional(),
   })
   // Tool usage must be enabled via config because we currently have no way to
   // track quota usage for them or enforce limits.