# Files in conversations (/en/docs/developers/model/guides/files)



See the [file service](/en/docs/developers/file) for uploads, expiry, downloads and deletion.

Use `file_id` to reference images, text files or PDFs uploaded with `purpose=user_data`. Pass `{"type":"file","file":{"file_id":"ID returned by upload"}}` in message content. Acceptance does not guarantee that the selected model can read the content. Files uploaded for `vision` or `assistants` are not used as image input here.

Verified on Develop on 2026-10-13: `qwen3.8-flash` extracted the requested phrase from both text and PDF files. For the same PDF, `deepseek-flash` returned HTTP 200 but reported conversion failure (`provider_rejected`) and stated that the file was not read. Switch to a verified model or place extracted text directly in the message. HTTP 200 alone does not prove the file was read.

Image input still requires an image-capable model; this example uses `qwen3-vl-plus`.

Install `openai` and set `PRAXIS_API_KEY`. For Develop, set `PRAXIS_API_BASE=https://api-develop.deepraxis.ai`. The example uploads a tiny PNG, asks for a description and deletes the file.

```python
import base64
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["PRAXIS_API_KEY"],
    base_url=os.environ.get("PRAXIS_API_BASE", "https://api.deepraxis.com") + "/openai/v1",
)
image = base64.b64decode(
    "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"
)
file = client.files.create(file=("pixel.png", image, "image/png"), purpose="user_data")
try:
    response = client.chat.completions.create(
        model="qwen3-vl-plus",
        max_tokens=80,
        messages=[{"role": "user", "content": [
            {"type": "text", "text": "Describe this image in one sentence."},
            {"type": "file", "file": {"file_id": file.id}},
        ]}],
    )
    print(response.choices[0].message.content)
finally:
    client.files.delete(file.id)
```

Expected: a brief image description; the file is deleted when the request finishes.