When using the AIProxy with Braintrust logging, there are cases where the LLM call can fail / be invalid. In all of these cases the span seems to never update, i.e. go into an error state.
Replication steps:
const client = new OpenAI({
baseURL: "https://api.braintrust.dev/v1/proxy",
apiKey: process.env.BRAINTRUST_API_KEY
});
const response = await client.chat.completions.create({
model: "gpt-4o-mini",
messages: [{ role: "user", content: "What is a proxy?" }],
timeout: 60000 // <------------ Unsupported Param
});
This snippet throws the following error: "Error: 400 Invalid JSON payload received. Unknown name \\"timeout\\" at 'generation_config': Cannot find field.
But the span in Braintrust is never updated. It shows no output and (no error as well)

Also in some cases, the span never ends:

When using the AIProxy with Braintrust logging, there are cases where the LLM call can fail / be invalid. In all of these cases the span seems to never update, i.e. go into an
errorstate.Replication steps:
This snippet throws the following error:
"Error: 400 Invalid JSON payload received. Unknown name \\"timeout\\" at 'generation_config': Cannot find field.But the span in Braintrust is never updated. It shows no
outputand (noerroras well)Also in some cases, the span never ends: