Pass the model in, allow changing it

Author: matt-aitken

apps/webapp/app/env.server.ts

@@ -975,6 +975,9 @@ const EnvironmentSchema = z.object({
   BULK_ACTION_BATCH_SIZE: z.coerce.number().int().default(100),
   BULK_ACTION_BATCH_DELAY_MS: z.coerce.number().int().default(200),
   BULK_ACTION_SUBBATCH_CONCURRENCY: z.coerce.number().int().default(5),
+
+  // AI Run Filter
+  AI_RUN_FILTER_MODEL: z.string().optional(),
 });
 
 export type Environment = z.infer<typeof EnvironmentSchema>;

apps/webapp/app/routes/resources.orgs.$organizationSlug.projects.$projectParam.env.$envParam.runs.ai-filter.tsx

@@ -1,3 +1,4 @@
+import { openai } from "@ai-sdk/openai";
 import { type ActionFunctionArgs, json } from "@remix-run/server-runtime";
 import { tryCatch } from "@trigger.dev/core";
 import { z } from "zod";
@@ -139,12 +140,15 @@ export async function action({ request, params }: ActionFunctionArgs) {
     );
   }
 
-  const service = new AIRunFilterService({
-    queryTags,
-    queryVersions,
-    queryQueues,
-    queryTasks,
-  });
+  const service = new AIRunFilterService(
+    {
+      queryTags,
+      queryVersions,
+      queryQueues,
+      queryTasks,
+    },
+    openai(env.AI_RUN_FILTER_MODEL ?? "gpt-4o-mini")
+  );
 
   const [error, result] = await tryCatch(service.call(text, environment.id));
   if (error) {

apps/webapp/app/v3/services/aiRunFilterService.server.ts

@@ -1,6 +1,6 @@
 import { openai } from "@ai-sdk/openai";
 import { type TaskTriggerSource } from "@trigger.dev/database";
-import { generateText, Output, tool } from "ai";
+import { generateText, LanguageModelV1, Output, tool } from "ai";
 import { z } from "zod";
 import { TaskRunListSearchFilters } from "~/components/runs/v3/RunFilters";
 import { logger } from "~/services/logger.server";
@@ -79,13 +79,14 @@ export class AIRunFilterService {
       queryVersions: QueryVersions;
       queryQueues: QueryQueues;
       queryTasks: QueryTasks;
-    }
+    },
+    private readonly model: LanguageModelV1 = openai("gpt-4o-mini")
   ) {}
 
   async call(text: string, environmentId: string): Promise<AIFilterResult> {
     try {
       const result = await generateText({
-        model: openai("gpt-4o-mini"),
+        model: this.model,
         experimental_output: Output.object({ schema: AIFilterResponseSchema }),
         tools: {
           lookupTags: tool({

apps/webapp/evals/aiRunFilter.eval.ts

@@ -8,6 +8,8 @@ import {
   type QueryVersions,
 } from "~/v3/services/aiRunFilterService.server";
 import dotenv from "dotenv";
+import { traceAISDKModel } from "evalite/ai-sdk";
+import { openai } from "@ai-sdk/openai";
 
 dotenv.config({ path: "../../.env" });
 
@@ -232,12 +234,15 @@ evalite("AI Run Filter", {
     ];
   },
   task: async (input) => {
-    const service = new AIRunFilterService({
-      queryTags,
-      queryVersions,
-      queryQueues,
-      queryTasks,
-    });
+    const service = new AIRunFilterService(
+      {
+        queryTags,
+        queryVersions,
+        queryQueues,
+        queryTasks,
+      },
+      traceAISDKModel(openai("gpt-4o-mini"))
+    );
 
     const result = await service.call(input, "123456");
     return JSON.stringify(result);