Create Chat Completion (OpenAI Chat Completions API Compatible) | Caesar

Files (Caesar Extension)

This endpoint supports file attachments using a Caesar extension to the Chat Completions message format. To include files in your research:

Upload files first via POST /research/files using multipart/form-data
Reference the returned UUID in a message content part of type file

Files are referenced using file.fileData, which must contain a Caesar file UUID (not raw bytes, base64, or a URL). The server loads the stored file by that ID and extracts usable text for the research.

Example content part:

1 {
2   "type": "file",
3   "file": {
4     "filename": "document.pdf",
5     "fileData": "9c6f8b1a-2a4f-4a35-86b9-0d0b5e25d5e5"
6   }
7 }

Note: The legacy /compat/completions endpoint does not accept files.

OpenAI Chat Completions API compatible endpoint. Creates a research job using the last user message and returns the result in OpenAI chat completion format. Supports both streaming and non-streaming responses. ## Files (Caesar Extension) This endpoint supports file attachments using a Caesar extension to the Chat Completions message format. To include files in your research: 1. **Upload files first** via `POST /research/files` using multipart/form-data 2. **Reference the returned UUID** in a message content part of type `file` Files are referenced using `file.fileData`, which must contain a **Caesar file UUID** (not raw bytes, base64, or a URL). The server loads the stored file by that ID and extracts usable text for the research. **Example content part:** ```json { "type": "file", "file": { "filename": "document.pdf", "fileData": "9c6f8b1a-2a4f-4a35-86b9-0d0b5e25d5e5" } } ``` **Note:** The legacy `/compat/completions` endpoint does not accept files.

Authentication

AuthorizationBearer

Bearer authentication of the form Bearer <token>, where token is your auth token.

Request

This endpoint expects an object.

modelstringRequired

Model identifier (accepted for compatibility).

messageslist of objectsRequired

List of messages in the conversation.

max_tokensintegerOptional>=1

Maximum number of tokens to generate (accepted for compatibility).

temperaturedoubleOptional0-2Defaults to 1

Sampling temperature (accepted for compatibility).

top_pdoubleOptional0-1Defaults to 1

Nucleus sampling parameter (accepted for compatibility).

stopstring or list of stringsOptional

Sequences where the API will stop generating (accepted for compatibility).

seedintegerOptional

Random seed for deterministic sampling (accepted for compatibility).

streambooleanOptionalDefaults to false

Whether to stream partial progress via SSE.

stream_optionsobjectOptional

metadataobjectOptional

Research configuration parameters. When not provided, auto mode is used to determine optimal settings.

Response

Chat completion response (or SSE stream if stream=true).

idstring

Unique identifier for the chat completion.

objectenum

Object type.

Allowed values:

createdinteger

Unix timestamp of when the completion was created.

modelstring

Model used for the completion.

choiceslist of objects

List of chat completion choices.

usageobject

Errors

Files (Caesar Extension)

This endpoint supports file attachments using a Caesar extension to the Chat Completions message format. To include files in your research:

Upload files first via POST /research/files using multipart/form-data
Reference the returned UUID in a message content part of type file

Example content part:

1 {
2   "type": "file",
3   "file": {
4     "filename": "document.pdf",
5     "fileData": "9c6f8b1a-2a4f-4a35-86b9-0d0b5e25d5e5"
6   }
7 }

Note: The legacy /compat/completions endpoint does not accept files.

1	import requests
2
3	url = "https://api.caesar.xyz/compat/chat/completions"
4
5	payload = { "body": {
6	"model": "caesar-research",
7	"messages": [
8	{
9	"role": "user",
10	"content": "What are the latest breakthroughs in quantum computing for 2024?"
11	}
12	],
13	"max_tokens": 512,
14	"temperature": 0.7,
15	"top_p": 0.9,
16	"stop": ["
17
18	"],
19	"seed": 42,
20	"stream": False,
21	"metadata": {
22	"reasoning_loops": 4,
23	"source_timeout": 30,
24	"reasoning_mode": True,
25	"allow_early_exit": True,
26	"exclude_social": False,
27	"system_prompt": "You are a research assistant providing concise, accurate summaries."
28	}
29	} }
30	headers = {
31	"Authorization": "Bearer <token>",
32	"Content-Type": "application/json"
33	}
34
35	response = requests.post(url, json=payload, headers=headers)
36
37	print(response.json())

1	{
2	"id": "9c6f8b1a-2a4f-4a35-86b9-0d0b5e25d5e5",
3	"object": "chat.completion",
4	"created": 1700000000,
5	"model": "caesar-research",
6	"choices": [
7	{
8	"index": 0,
9	"message": {
10	"role": "assistant",
11	"content": "Quantum computing has seen significant advances in 2024, including improved qubit coherence times, scalable error correction techniques, and new quantum algorithms that enhance cryptography and optimization problems."
12	},
13	"finish_reason": "stop"
14	}
15	],
16	"usage": {
17	"prompt_tokens": 45,
18	"completion_tokens": 60,
19	"total_tokens": 105
20	}
21	}

1	{
2	"type": "file",
3	"file": {
4	"filename": "document.pdf",
5	"fileData": "9c6f8b1a-2a4f-4a35-86b9-0d0b5e25d5e5"
6	}
7	}