chore: use 3.5-sonnet

Author: deloreyj

package.json

@@ -26,7 +26,8 @@
 		"test": "vitest run --passWithNoTests",
 		"fix:format": "prettier . --write",
 		"test:watch": "vitest",
-		"eval:ci": "run-turbo eval:ci"
+		"eval:ci": "run-turbo eval:ci",
+		"eval:dev": "run-turbo eval:dev"
 	},
 	"devDependencies": {
 		"@changesets/cli": "2.28.1",

packages/eval-tools/src/test-models.ts

@@ -56,7 +56,7 @@ function getWorkersAiModel(modelName: AiTextGenerationModels) {
 export const eachModel = describe.each([
 	getOpenAiModel('gpt-4o'),
 	getOpenAiModel('gpt-4o-mini'),
-	getAnthropicModel('claude-3-7-sonnet-latest'),
+	getAnthropicModel('claude-3-5-sonnet-latest'),
 	// llama 3 is somewhat inconsistent
 	//getWorkersAiModel("@cf/meta/llama-3.3-70b-instruct-fp8-fast")
 	// Currently llama 4 is having issues with tool calling