Completions (legacy)¶
Older code that sends a prompt to /v1/completions works through the gateway unchanged, so it's checked, charged and
recorded like everything else while you plan its move to Chat Completions.
What it is¶
POST /v1/completions, OpenAI's original text-completion endpoint: a prompt in, text out. OpenAI itself marks it
as legacy. The gateway passes it on unchanged to OpenAI and Azure OpenAI, and to a provider added with the OpenAI
kind of API that still offers it. It isn't translated: a call to a Claude or Gemini model gets a 400.
The two settings¶
The same as for the OpenAI SDK: base address https://<your-gateway>/v1 and your Vidai key. See
OpenAI SDK for Python.
A minimal call¶
from openai import OpenAI
client = OpenAI() # reads OPENAI_BASE_URL and OPENAI_API_KEY
completion = client.completions.create(
model="gpt-3.5-turbo-instruct",
prompt="Say hello in Dutch:",
max_tokens=20,
)
print(completion.choices[0].text)
import OpenAI from "openai";
const client = new OpenAI();
const completion = await client.completions.create({
model: "gpt-3.5-turbo-instruct",
prompt: "Say hello in Dutch:",
max_tokens: 20,
});
console.log(completion.choices[0].text);
curl https://<your-gateway>/v1/completions \
-H "Authorization: Bearer $OPENAI_API_KEY" -H "Content-Type: application/json" \
-d '{"model": "gpt-3.5-turbo-instruct", "prompt": "Say hello in Dutch:", "max_tokens": 20}'
The model name is an example: use one your organisation approved under Models that the provider serves on this endpoint.
Streaming¶
stream=True works as OpenAI documents it; the chunks are the provider's own, passed through. As with chat, a
fallback model is tried only before the first byte.
Reading what the gateway did¶
Use client.completions.with_raw_response.create(...) (Python) or .withResponse() (TypeScript) and read
x-vidai-request-id and x-vidai-model, as shown on Chat Completions.
Errors¶
The codes are those of Chat Completions. Specific to this endpoint: a model on a provider that doesn't speak the OpenAI API (Claude, Gemini, Bedrock) gets a 400, and nothing is sent.
What goes through the gateway¶
| Feature | Through the gateway |
|---|---|
completions.create, streaming |
Yes, to OpenAI and Azure OpenAI |
| Claude, Gemini and Bedrock models | No: a 400. Use Chat Completions |
Check it works in a minute¶
Run the curl above. A short answer means the address, key and model are right; the call shows in Request logs within seconds.
Good to know¶
- On this endpoint, a guardrail set to mask blocks the call instead.
- Moving to Chat Completions opens every provider to the same code: a
promptbecomesmessages=[{"role": "user", "content": prompt}].