oai responses api, tool call retries

server/src/llm/streaming.ts

@@ -1,7 +1,12 @@
 import { performance } from "node:perf_hooks";
 import { prisma } from "../db.js";
 import { anthropicClient, openaiClient, xaiClient } from "./providers.js";
-import { buildToolLogMessageData, runToolAwareOpenAIChatStream, type ToolExecutionEvent } from "./chat-tools.js";
+import {
+  buildToolLogMessageData,
+  runToolAwareChatCompletionsStream,
+  runToolAwareOpenAIChatStream,
+  type ToolExecutionEvent,
+} from "./chat-tools.js";
 import { buildAnthropicConversationMessage, getAnthropicSystemPrompt } from "./message-content.js";
 import type { MultiplexRequest, Provider } from "./types.js";
 
@@ -58,18 +63,33 @@ export async function* runMultiplexStream(req: MultiplexRequest): AsyncGenerator
   try {
     if (req.provider === "openai" || req.provider === "xai") {
       const client = req.provider === "openai" ? openaiClient() : xaiClient();
-      for await (const ev of runToolAwareOpenAIChatStream({
-        client,
-        model: req.model,
-        messages: req.messages,
-        temperature: req.temperature,
-        maxTokens: req.maxTokens,
-        logContext: {
-          provider: req.provider,
-          model: req.model,
-          chatId,
-        },
-      })) {
+      const streamEvents =
+        req.provider === "openai"
+          ? runToolAwareOpenAIChatStream({
+              client,
+              model: req.model,
+              messages: req.messages,
+              temperature: req.temperature,
+              maxTokens: req.maxTokens,
+              logContext: {
+                provider: req.provider,
+                model: req.model,
+                chatId,
+              },
+            })
+          : runToolAwareChatCompletionsStream({
+              client,
+              model: req.model,
+              messages: req.messages,
+              temperature: req.temperature,
+              maxTokens: req.maxTokens,
+              logContext: {
+                provider: req.provider,
+                model: req.model,
+                chatId,
+              },
+            });
+      for await (const ev of streamEvents) {
         if (ev.type === "delta") {
           text += ev.text;
           yield { type: "delta", text: ev.text };