Chat Completions

POST https://revidapi.com/v1/chat/completions

OpenAI-compatible. Any library that speaks to OpenAI works here — point base_url at https://revidapi.com/v1.

Auth

Send either header:

Authorization: Bearer sk_YOUR_KEY
X-API-Key: sk_YOUR_KEY

Parameters

Field	Type	Required	Description
model	string	Yes	Model ID, e.g. claude-sonnet-5. See models
messages	array	Yes	{role, content} items. role is system, user or assistant
stream	bool	No	true to receive SSE chunks
max_tokens	int	No	Caps the answer length — the surest way to avoid paying for an unexpectedly long reply
temperature	number	No	0–2
top_p	number	No	0–1
tools	array	No	OpenAI-style function definitions
tool_choice	string/object	No	auto, none, or a named function
parallel_tool_calls	bool	No	Allow several tool calls per turn

Basic example

curl https://revidapi.com/v1/chat/completions \
  -H "Authorization: Bearer sk_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "max_tokens": 1024,
    "messages": [
      {"role": "system", "content": "Answer briefly."},
      {"role": "user", "content": "Explain HTTP 429 in two sentences."}
    ]
  }'

Response:

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "model": "claude-sonnet-5",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {"role": "assistant", "content": "429 means..."}
    }
  ],
  "usage": {"prompt_tokens": 42, "completion_tokens": 58, "total_tokens": 100}
}

Streaming

Set "stream": true. The server returns Server-Sent Events, each line prefixed with data:, terminated by data: [DONE].

from openai import OpenAI

client = OpenAI(api_key="sk_YOUR_KEY",
                base_url="https://revidapi.com/v1")

for chunk in client.chat.completions.create(
    model="claude-sonnet-5",
    messages=[{"role": "user", "content": "Tell me a short story."}],
    max_tokens=800,
    stream=True,
):
    piece = chunk.choices[0].delta.content
    if piece:
        print(piece, end="", flush=True)

Function calling

curl https://revidapi.com/v1/chat/completions \
  -H "Authorization: Bearer sk_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "What is the weather in Hanoi?"}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "get_weather",
        "description": "Current weather for a city",
        "parameters": {
          "type": "object",
          "properties": {"city": {"type": "string"}},
          "required": ["city"]
        }
      }
    }]
  }'

The model returns tool_calls. Run the function on your side, then send the result back as a role: "tool" message, exactly as with OpenAI.

Not every model supports this. The Tools column in models says which do.

Images

For models with ✅ in the Vision column:

{
  "model": "claude-sonnet-5",
  "max_tokens": 1024,
  "messages": [{
    "role": "user",
    "content": [
      {"type": "text", "text": "What is in this image?"},
      {"type": "image_url",
       "image_url": {"url": "data:image/jpeg;base64,/9j/4AAQ..."}}
    ]
  }]
}

Error codes

Code	Meaning	What to do
400	Bad body — missing model, empty messages, malformed JSON	Fix the request
401	Missing or invalid key	Check the header
402	Not enough credits	Top up
404	Model not on sale	Call GET /v1/models for valid names
429	Too many concurrent calls	Back off and retry
503	Upstream model failing	Retry later, or pick another model

Errors cost nothing. Any call that does not return an answer is refunded automatically. You can verify this in GET /v1/credits — the ledger records every charge and every refund.

We never substitute models on your behalf. If you ask for claude-opus-5 and its upstream is failing, we return an error and refund — we do not quietly answer with a cheaper model.