Key overview

OpenAI embedding capability

OpenAI embedding capability should not be guessed. A useful read-only probe separates provider identity, models, balance, limits, and risk signals.

Provider

OpenAI

global provider route

Check dimension

embedding capability

embedding-capability

Catalog models

GPT-3.5 Turbo · GPT-3.5 Turbo (older v0613) · GPT-3.5 Turbo 16k

Privacy

5 min

Read-only check. Detection data burns after 5 minutes.

Check summary

Provider

OpenAI

openai

Check dimension

embedding capability

embedding-capability

Read-only check. Detection data burns after 5 minutes.

The result should be evidence-based, not only pass/fail, so it can feed admin, monitoring, and API layers later.

Why it matters

embedding capability decides whether a key can move from “connects” to “manageable, billable, and operable”.

Confirm whether the key really belongs to OpenAI, not only a compatibility layer, proxy, or forwarding route.
Record evidence for embedding capability: confirmed, partially visible, invisible, or not exposed by the provider.

The result should be evidence-based, not only pass/fail, so it can feed admin, monitoring, and API layers later.

If evidence is strong enough, move to pricing, listing, monitoring, and alerts; if not, mark the provider interface as not exposed.