OpenAI Files and Batches

Endpoint

import os
from openai import OpenAI

base_url = os.environ["MODROUTER_BASE_URL"].rstrip("/")
api_key = os.environ["MODROUTER_API_KEY"]

client = OpenAI(
    base_url=f"{base_url}/openai",
    api_key=api_key,
)

Files

ModRouter registers OpenAI-compatible file routes under /openai. With base_url=f"{base_url}/openai", OpenAI SDK calls use the first route form below; clients configured with /openai/v1 use the second.

SDK operation	ModRouter route
upload	`POST /openai/files` or `POST /openai/v1/files`
list	`GET /openai/files` or `GET /openai/v1/files`
retrieve	`GET /openai/files/{file_id}` or `GET /openai/v1/files/{file_id}`
delete	`DELETE /openai/files/{file_id}` or `DELETE /openai/v1/files/{file_id}`
content	`GET /openai/files/{file_id}/content` or `GET /openai/v1/files/{file_id}/content`

Example:

uploaded = client.files.create(
    file=open("requests.jsonl", "rb"),
    purpose="batch",
)

print(uploaded.id)

Batches

ModRouter registers OpenAI-compatible batch routes under /openai:

SDK operation	ModRouter route
create	`POST /openai/batches` or `POST /openai/v1/batches`
list	`GET /openai/batches` or `GET /openai/v1/batches`
retrieve	`GET /openai/batches/{batch_id}` or `GET /openai/v1/batches/{batch_id}`
cancel	`POST /openai/batches/{batch_id}/cancel` or `POST /openai/v1/batches/{batch_id}/cancel`

Example:

batch = client.batches.create(
    input_file_id=uploaded.id,
    endpoint="/v1/chat/completions",
    completion_window="24h",
)

print(batch.id)

Containers

ModRouter also exposes OpenAI-compatible container and container-file routes under /openai/containers and /openai/v1/containers. Use these only with providers configured for container operations. Unsupported providers return a normal ModRouter provider error.

Cross-Provider Notes

OpenAI SDK routes can target other providers when ModRouter can translate the operation. Include provider information through the model name or operation-specific request fields supported by ModRouter. For provider-native batch or file semantics, use the matching provider integration:

Anthropic inline message batches: /anthropic
Google GenAI files and batches: /genai
Bedrock model invocation jobs and S3-compatible file routes: /bedrock

​Endpoint

​Files

​Batches

​Containers

​Cross-Provider Notes

Endpoint

Files

Batches

Containers

Cross-Provider Notes