gpt5pro

src/config.ts

@@ -963,6 +963,6 @@ function parseCsv(val: string): string[] {
 
 function getDefaultModelFamilies(): ModelFamily[] {
   return MODEL_FAMILIES.filter(
-    (f) => !f.includes("o1-pro") && !f.includes("o3-pro")
+    (f) => !f.includes("o1-pro") && !f.includes("o3-pro") && !f.includes("gpt5-pro")
   ) as ModelFamily[];
 }

src/info-page.ts

@@ -44,6 +44,7 @@ const MODEL_FAMILY_FRIENDLY_NAME: { [f in ModelFamily]: string } = {
   gpt5: "GPT-5",
   "gpt5-mini": "GPT-5 Mini",
   "gpt5-nano": "GPT-5 Nano",
+  "gpt5-pro": "GPT-5 Pro",
   "gpt5-chat-latest": "GPT-5 Chat Latest",
   gpt45: "GPT-4.5",
   o1: "OpenAI o1",
@@ -85,6 +86,7 @@ const MODEL_FAMILY_FRIENDLY_NAME: { [f in ModelFamily]: string } = {
   "azure-gpt5": "Azure GPT-5",
   "azure-gpt5-mini": "Azure GPT-5 Mini",
   "azure-gpt5-nano": "Azure GPT-5 Nano",
+  "azure-gpt5-pro": "GPT-5 Pro (Azure)",
   "azure-gpt5-chat-latest": "Azure GPT-5 Chat Latest",
   "azure-o1": "Azure o1",
   "azure-o1-mini": "Azure o1 mini",
@@ -402,3 +404,4 @@ if (config.enableInfoPageLogin) {
 /*  ─── Removed the public /status route :  simply not added ─── */
 
 export { infoPageRouter };
+

src/proxy/middleware/request/index.ts

@@ -16,6 +16,7 @@ export { transformOutboundPayload } from "./preprocessors/transform-outbound-pay
 export { validateContextSize } from "./preprocessors/validate-context-size";
 export { validateModelFamily } from "./preprocessors/validate-model-family";
 export { validateVision } from "./preprocessors/validate-vision";
+export { validateStreaming } from "./preprocessors/validate-streaming";
 export { extractQwenExtraBody } from "./preprocessors/extract-qwen-extra-body";
 
 // Proxy request mutators (runs every time request is dequeued, before proxying, usually for auth/signing)

src/proxy/middleware/request/preprocessor-factory.ts

@@ -12,6 +12,7 @@ import {
   validateContextSize,
   validateModelFamily,
   validateVision,
+  validateStreaming,
   applyQuotaLimits,
 } from ".";
 
@@ -57,6 +58,7 @@ export const createPreprocessorMiddleware = (
     ...(afterTransform ?? []),
     validateContextSize,
     validateVision,
+    validateStreaming,
     validateModelFamily,
     applyQuotaLimits,
   ];

src/proxy/middleware/request/preprocessors/transform-outbound-payload.ts

@@ -102,6 +102,19 @@ function applyOpenAIResponsesTransform(req: Request): void {
       }
     }
 
+    // Handle reasoning_effort for models that require it
+    const model = req.body.model || "";
+    const isO3Pro = model === "o3-pro" || model.startsWith("o3-pro-");
+    const isGpt5Pro = model === "gpt-5-pro" || model.startsWith("gpt-5-pro-");
+    
+    // o3-pro and gpt-5-pro default to and only support "high" reasoning effort
+    if (isO3Pro || isGpt5Pro) {
+      if (!req.body.reasoning_effort || req.body.reasoning_effort !== "high") {
+        req.body.reasoning_effort = "high";
+        req.log.info({ model, reasoning_effort: "high" }, "Set reasoning_effort to 'high' (required for this model)");
+      }
+    }
+
     req.log.info({
       originalModel: originalBody.model,
       newFormat: "openai-responses"

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

@@ -102,6 +102,8 @@ export const validateContextSize: RequestPreprocessor = async (req) => {
   } else if (model.match(/^o1-pro(-\d{4}-\d{2}-\d{2})?$/)) {
     modelMax = 200000;
   } else if (model.match(/^o3-pro(-\d{4}-\d{2}-\d{2})?$/)) {
+    modelMax = 400000;
+  } else if (model.match(/^gpt-5-pro(-\d{4}-\d{2}-\d{2})?$/)) {
     modelMax = 200000;
   } else if (model.match(/^o1-preview(-\d{4}-\d{2}-\d{2})?$/)) {
     modelMax = 128000;

src/proxy/middleware/request/preprocessors/validate-streaming.ts

@@ -0,0 +1,30 @@
+import { BadRequestError } from "../../../../shared/errors";
+import { RequestPreprocessor } from "../index";
+
+/**
+ * Throws an error if streaming is requested for models that don't support it
+ * (o3-pro, o1-pro, gpt-5-pro and their variants).
+ */
+export const validateStreaming: RequestPreprocessor = (req) => {
+  const { model, stream } = req.body;
+  
+  // Check if streaming is enabled
+  const isStreaming = stream === "true" || stream === true;
+  if (!isStreaming) {
+    return;
+  }
+  
+  // Check if model is one of the non-streaming models
+  const modelStr = String(model).toLowerCase();
+  const nonStreamingModels = ["o3-pro", "o1-pro", "gpt-5-pro"];
+  
+  const isNonStreamingModel = nonStreamingModels.some((modelName) =>
+    modelStr.includes(modelName)
+  );
+  
+  if (isNonStreamingModel) {
+    throw new BadRequestError(
+      "Streaming is not supported for this model. The dev is too lazy to implement it. Please set stream=false."
+    );
+  }
+};

src/proxy/openai.ts

@@ -371,6 +371,7 @@ const setupChunkedTransfer: RequestHandler = (req, res, next) => {
 function shouldUseResponsesApi(model: string): boolean {
   return model === "o1-pro" || model.startsWith("o1-pro") ||
          model === "o3-pro" || model.startsWith("o3-pro") ||
+         model === "gpt-5-pro" || model.startsWith("gpt-5-pro") ||
          model === "codex-mini-latest" || model.startsWith("codex-mini") ||
          model === "gpt-5-codex-latest" || model.startsWith("gpt-5-codex");
 }

src/service-info.ts

@@ -56,6 +56,7 @@ const MODEL_FAMILY_ORDER: ModelFamily[] = [
   "gpt5-mini",
   "gpt5-nano",
   "gpt5-chat-latest",
+  "gpt5-pro",
   "o1",
   "o1-mini",
   "o1-pro",
@@ -80,6 +81,7 @@ const MODEL_FAMILY_ORDER: ModelFamily[] = [
   "azure-gpt5-mini",
   "azure-gpt5-nano",
   "azure-gpt5-chat-latest",
+  "azure-gpt5-pro",
   "azure-o1",
   "azure-o1-mini",
   "azure-o1-pro",

src/shared/models.ts

@@ -36,6 +36,7 @@ export type OpenAIModelFamily =
   | "gpt5-mini"
   | "gpt5-nano"
   | "gpt5-chat-latest"
+  | "gpt5-pro"
   | "o1"
   | "o1-mini"
   | "o1-pro"
@@ -104,6 +105,7 @@ export const MODEL_FAMILIES = (<A extends readonly ModelFamily[]>(
   "gpt5-mini",
   "gpt5-nano",
   "gpt5-chat-latest",
+  "gpt5-pro",
   "o1",
   "o1-mini",
   "o1-pro",
@@ -144,6 +146,7 @@ export const MODEL_FAMILIES = (<A extends readonly ModelFamily[]>(
   "azure-gpt5-mini",
   "azure-gpt5-nano",
   "azure-gpt5-chat-latest",
+  "azure-gpt5-pro",
   "azure-dall-e",
   "azure-o1",
   "azure-o1-mini",
@@ -196,6 +199,7 @@ export const MODEL_FAMILY_SERVICE: {
   "gpt5-mini": "openai",
   "gpt5-nano": "openai",
   "gpt5-chat-latest": "openai",
+  "gpt5-pro": "openai",
   "o1": "openai",
   "o1-mini": "openai",
   "o1-pro": "openai",
@@ -229,6 +233,7 @@ export const MODEL_FAMILY_SERVICE: {
   "azure-gpt5-mini": "azure",
   "azure-gpt5-nano": "azure",
   "azure-gpt5-chat-latest": "azure",
+  "azure-gpt5-pro": "azure",
   "azure-dall-e": "azure",
   "azure-o1": "azure",
   "azure-o1-mini": "azure",
@@ -256,6 +261,7 @@ export const OPENAI_MODEL_FAMILY_MAP: { [regex: string]: OpenAIModelFamily } = {
   "^gpt-5-mini(-\\d{4}-\\d{2}-\\d{2})?$": "gpt5-mini",
   "^gpt-5-nano(-\\d{4}-\\d{2}-\\d{2})?$": "gpt5-nano",
   "^gpt-5-chat-latest(-\\d{4}-\\d{2}-\\d{2})?$": "gpt5-chat-latest",
+  "^gpt-5-pro(-\\d{4}-\\d{2}-\\d{2})?$": "gpt5-pro",
   "^gpt-4\\.5(-preview)?(-\\d{4}-\\d{2}-\\d{2})?$": "gpt45",
   "^gpt-4\\.1(-\\d{4}-\\d{2}-\\d{2})?$": "gpt41",
   "^gpt-4\\.1-mini(-\\d{4}-\\d{2}-\\d{2})?$": "gpt41-mini",

src/shared/stats.ts

@@ -20,6 +20,8 @@ const MODEL_PRICING: Record<ModelFamily, { input: number; output: number } | und
   "azure-gpt5-nano": { input: 0.05, output: 0.40 },
   "gpt5-chat-latest": { input: 1.25, output: 10.00 },
   "azure-gpt5-chat-latest": { input: 1.25, output: 10.00 },
+  "gpt5-pro": { input: 15.00, output: 120.00 },
+  "azure-gpt5-pro": { input: 15.00, output: 120.00 },
   "gpt45": { input: 75.00, output: 150.00 }, // Example, needs verification if this model family is still current with this pricing
   "azure-gpt45": { input: 75.00, output: 150.00 }, // Example, needs verification
   "gpt4o": { input: 2.50, output: 10.00 },