Use the Responses, embeddings, rerank and moderations APIs

Proxium serves four more OpenAI APIs with the same base URL and virtual key as chat:

Route	What it does	Routing type	Stream
POST /v1/embeddings	Turns text into vectors	Embeddings	No
POST /v1/responses	The OpenAI Responses API	Text	Yes
POST /v1/rerank	Orders documents by how well they match a query	Text	No
POST /v1/moderations	Checks text against a content policy	Text	No

Proxium sends your request body to the same path at the vendor, for example <base URL of the vendor>/embeddings. It sends the answer of the vendor back without change.

Your appPOST /v1/embeddings

→↓

Proxiumchecks the key and the limits, picks the model

→↓

The vendorPOST <its base URL>/embeddings

→↓

Your appgets the answer of the vendor, unchanged

warning

The vendor must serve that path. For example, a Groq vendor serves chat. A rerank call that goes to it fails, and Proxium tries the next model of the chain.

Make embeddings

1. Add a vendor that serves embeddings, for example the OpenAI (embeddings) preset. See Add vendors and your own keys.
2. Get the model id from GET /v1/models, for example t.acme.openai-embed/text-embedding-3-small.
3. Send the id in model:

Python

result = client.embeddings.create(
    model="t.acme.openai-embed/text-embedding-3-small",
    input=["The first text.", "The second text."],
)
print(len(result.data), len(result.data[0].embedding))

curl

Terminal

curl https://proxium.tech/v1/embeddings \
  -H "Authorization: Bearer $PROXIUM_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "t.acme.openai-embed/text-embedding-3-small", "input": ["The first text."]}'

Proxium does not store or search the vectors. It records the tokens and the cost of the call.

To send a tier name in place of the id, set a chain for the tier embed in Routing › Embeddings routing.

Use the Responses API

Send the call to a vendor that serves the Responses API, such as OpenAI. Set stream to true to get server-sent events.

response = client.responses.create(
    model="t.acme.openai/gpt-4o-mini",
    input="Write one sentence about queues.",
)
print(response.output_text)

Rerank and moderations

Send the body that your vendor documents. Proxium needs only the model field:

Terminal

curl https://proxium.tech/v1/rerank \
  -H "Authorization: Bearer $PROXIUM_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "t.acme.my-reranker/bge-reranker-v2-m3", "query": "refund policy", "documents": ["Refunds take 5 days.", "We ship worldwide."]}'

Here my-reranker is a server that you added with A server I run myself, with the request type Text.

What is the same as chat, and what is not

	Chat and Messages	These four routes
Virtual key, ceilings, cost record	Yes	Yes
Routing and failover	Yes	Yes
Row in the Attempt log	Yes	Yes
Response cache	Yes	No. Each call goes to a vendor
x-proxium-memory: recall	Adds memories	No effect. /v1/responses keeps the call in memory as write
A call without model	400 bad_json	400 missing_model

Related pages

• Route requests to models: set the chain of a tier.
• API reference: the body of each route.