Skip to content

Chat & General Inference

Before You Run Examples

All examples in this page assume environment variables are available. For convenience, create a .env file in the repository root (you can copy from .env.example) with:

TUDELFT_TULIP_BASE_URL=https://api.tulip.tudelft.nl
TUDELFT_TULIP_API_KEY=YOUR_API_KEY
TUDELFT_TULIP_MODEL_ALIAS=chat
TUDELFT_TULIP_CAPABILITY=chat

Set TUDELFT_TULIP_BASE_URL to host root only (for example https://api.tulip.tudelft.nl). The Python helper normalizes it to a capability-specific /v1 endpoint automatically.

What these examples assume

  • The Python snippets on this page assume a local helper file common.py that loads environment variables and exports shared constants.
  • If you cloned this repository, examples/common.py is already present and the scripts run as shown below.
  • If you have not cloned the repo and are working with Python, expand the helper snippet below and place common.py next to the example you want to run.
  • Python run commands use pixi run python ...; dependencies come from pixi.toml.
  • Bash examples pretty-print JSON using jq, so you may need to install it.
  • TUDELFT_TULIP_CAPABILITY chooses the path segment (chat, code, embed).
  • TUDELFT_TULIP_MODEL_ALIAS chooses the deployed model alias.
If you have not cloned the repo and are working with Python, create common.py next to the example
common.py
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
import os

from dotenv import load_dotenv

load_dotenv()  # loads .env if present

TUDELFT_TULIP_BASE_URL = os.getenv(
    "TUDELFT_TULIP_BASE_URL", "https://api.tulip.tudelft.nl"
)
TUDELFT_TULIP_CAPABILITY = os.getenv("TUDELFT_TULIP_CAPABILITY", "chat")
TUDELFT_TULIP_API_KEY = os.getenv("TUDELFT_TULIP_API_KEY", "DUMMY_KEY")
TUDELFT_TULIP_MODEL_ALIAS = os.getenv("TUDELFT_TULIP_MODEL_ALIAS", "chat")

# Normalize TUDELFT_TULIP_BASE_URL to capability/v1. Docs recommend host root only,
# but we also accept URLs that already include capability or capability/v1.
_base_url = TUDELFT_TULIP_BASE_URL.rstrip("/")
if _base_url.endswith("/v1"):
    TUDELFT_TULIP_BASE_URL = _base_url
elif _base_url.endswith(f"/{TUDELFT_TULIP_CAPABILITY}"):
    TUDELFT_TULIP_BASE_URL = f"{_base_url}/v1"
else:
    TUDELFT_TULIP_BASE_URL = f"{_base_url}/{TUDELFT_TULIP_CAPABILITY}/v1"

This tutorial covers two endpoints:

  • /v1/responses
  • /v1/chat/completions

Quick setup

  • Use curl-responses-basic.sh / python_responses_basic.py for /v1/responses, and curl-chat-completions.sh / python_chat_completions.py for /v1/chat/completions.

Responses API

Recommended approach

OpenAI recommends /v1/responses for new projects. It supports all capabilities (chat, code, embed) and provides a unified interface for text and non-text outputs. For more, seeOpenAI Responses API.

curl-responses-basic.sh
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
#!/bin/bash

SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
if [ -f "${SCRIPT_DIR}/../.env" ]; then
  source "${SCRIPT_DIR}/../.env"
fi

curl -s --fail --show-error ${TUDELFT_TULIP_BASE_URL}/${TUDELFT_TULIP_CAPABILITY}/v1/responses \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ${TUDELFT_TULIP_API_KEY}" \
  -d '{
    "model": "'"${TUDELFT_TULIP_MODEL_ALIAS}"'",
    "input": "Tell me a three sentence bedtime story about a unicorn."
  }' | jq .
Run: 
bash examples/curl-responses-basic.sh

SDK: OpenAI Python client (OpenAI)

python_responses_basic.py
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
from common import (
    TUDELFT_TULIP_API_KEY,
    TUDELFT_TULIP_BASE_URL,
    TUDELFT_TULIP_MODEL_ALIAS,
)
from openai import OpenAI

client = OpenAI(base_url=TUDELFT_TULIP_BASE_URL, api_key=TUDELFT_TULIP_API_KEY)
model = TUDELFT_TULIP_MODEL_ALIAS

response = client.responses.create(
    model=model,
    input="Tell me a three sentence bedtime story about a unicorn.",
)

print(response.output_text)
Run: 
pixi run python examples/python_responses_basic.py

Success looks like

  • Bash: curl-responses-basic.sh prints a valid JSON response object from /v1/responses.
  • Python: python_responses_basic.py prints extracted model text via response.output_text.

Chat Completions

Legacy endpoint (chat/completions only)

/v1/chat/completions is available for chat and code capabilities only. It does not support embeddings. For new projects or to work with all capabilities uniformly, prefer /v1/responses above. For more, seeOpenAI Chat Completions API.

curl-chat-completions.sh
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
#!/bin/bash

SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
if [ -f "${SCRIPT_DIR}/../.env" ]; then
  source "${SCRIPT_DIR}/../.env"
fi

curl -s --fail --show-error ${TUDELFT_TULIP_BASE_URL}/${TUDELFT_TULIP_CAPABILITY}/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ${TUDELFT_TULIP_API_KEY}" \
  -d '{
    "model": "'"${TUDELFT_TULIP_MODEL_ALIAS}"'",
    "max_tokens": 256,
    "temperature": 0.2,
    "messages": [
      { "role": "system", "content": "You are a helpful coding assistant for TU Delft." },
      { "role": "user", "content": "Write a tiny Python function that prints the TU Delft logo in ASCII." }
    ]
  }' | jq .
Run: 
bash examples/curl-chat-completions.sh

SDK: OpenAI Python client (OpenAI)

python_chat_completions.py
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
from common import (
    TUDELFT_TULIP_API_KEY,
    TUDELFT_TULIP_BASE_URL,
    TUDELFT_TULIP_MODEL_ALIAS,
)
from openai import OpenAI

client = OpenAI(base_url=TUDELFT_TULIP_BASE_URL, api_key=TUDELFT_TULIP_API_KEY)
model = TUDELFT_TULIP_MODEL_ALIAS

completion = client.chat.completions.create(
    model=model,
    messages=[
        {
            "role": "developer",
            "content": "You are a helpful coding assistant for TU Delft.",
        },
        {
            "role": "user",
            "content": (
                "Write a tiny Python function that prints the TU Delft logo in ASCII."
            ),
        },
    ],
)

print(completion.choices[0].message.content)
Run: 
pixi run python examples/python_chat_completions.py

Success looks like

  • Bash: curl-chat-completions.sh returns valid JSON from /v1/chat/completions.
  • Python: python_chat_completions.py prints assistant text from completion.choices[0].message.content.

Async & Streaming (Python)

Async responses

SDK: OpenAI Python async client (AsyncOpenAI)

python_responses_async.py
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
import asyncio

from common import (
    TUDELFT_TULIP_API_KEY,
    TUDELFT_TULIP_BASE_URL,
    TUDELFT_TULIP_MODEL_ALIAS,
)
from openai import AsyncOpenAI

client = AsyncOpenAI(base_url=TUDELFT_TULIP_BASE_URL, api_key=TUDELFT_TULIP_API_KEY)
model = TUDELFT_TULIP_MODEL_ALIAS


async def main() -> None:
    response = await client.responses.create(
        model=model, input="Explain disestablishmentarianism to a smart five year old."
    )
    print(response.output_text)


asyncio.run(main())

Run:

pixi run python examples/python_responses_async.py

Success looks like

  • Script finishes normally and prints text output.
  • No async transport or timeout errors for a basic request.

Streaming chat

Endpoint pattern: Chat completions with stream=True

Warning

Some deployments may not expose OpenAI Responses streaming events yet. This example uses chat-completions streaming for maximum compatibility. For more, seeOpenAI Streaming Responses guide.

python_responses_stream.py
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
from common import (
    TUDELFT_TULIP_API_KEY,
    TUDELFT_TULIP_BASE_URL,
    TUDELFT_TULIP_MODEL_ALIAS,
)
from openai import OpenAI

client = OpenAI(base_url=TUDELFT_TULIP_BASE_URL, api_key=TUDELFT_TULIP_API_KEY)
model = TUDELFT_TULIP_MODEL_ALIAS

stream = client.chat.completions.create(
    model=model,
    messages=[{"role": "user", "content": "Hello TULIP streaming test!"}],
    stream=True,
)

for chunk in stream:
    # Each chunk is a ChatCompletionChunk; content trickles in delta.
    if chunk.choices and chunk.choices[0].delta and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)
print()

Run:

pixi run python examples/python_responses_stream.py

Success looks like

  • Output appears progressively token-by-token or chunk-by-chunk.
  • Stream completes with no protocol or parsing errors.