Crafting Engaging Virtual Experiences

did you attach any default config to your api key?

if the config has an overriding model, it will override the param in your request

Ok, i dont think so. I dont have the option of using a config with the api key. Maybe thats because im still on the free plan? Still testing...

Request: {"mode":"completion","temperature":0.69999999999999996,"model":"gpt-4o"}

Response: "object":"chat.completion","created":1732085564,"model":"gpt-3.5-turbo-0125",

That's strange, can you attach your code snippet/http cURL?

url -X POST "https://api.portkey.ai/v1/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "x-portkey-provider: openai" \
-H "x-portkey-virtual-key: YOUR_OPENAI_API_KEY" \
-H "x-portkey-trace-id: assistant" \
-d '{
"config": {
"model": "gpt-4o",
"mode": "completion",
"temperature": 0.7
},
"messages": [
{
"role": "user",
"content": "System: prompt"
}
],
"trace": {
"user_id": "assistant",
"metadata": {
"app_version": "1.0",
"feature": "chat"
}
}
}'

ahh, you need to pass the config as a header, not as part of the body, to achieve the effect you're trying to get, you should do

Cool, thanks. Will try that 🙂

and completion is not a supported mode
mode is for routing strategy

checkout the supported json schema here https://portkey.ai/docs/api-reference/inference-api/config-object#json-schema