Inference API Quickstart

The Telnyx Inference API accepts OpenAI-compatible chat completion requests. Use a Telnyx API key and the SDK base URL https://api.telnyx.com/v2/ai/openai.

Prerequisites

A Telnyx account.
An API v2 key from API Keys in Mission Control Portal. See Create and manage API keys for key setup.
cURL for the shell example, or Python 3.10+ for the Python example.

1. Set your API key

Replace the placeholder with the Telnyx API key. Run this in the same terminal used for the requests below:

export TELNYX_API_KEY="YOUR_TELNYX_API_KEY"

$env:TELNYX_API_KEY="YOUR_TELNYX_API_KEY"

Keep the key on the server or in a local environment variable. Do not commit it to source control or include it in browser or mobile application code.

2. Send a chat completion

These examples use zai-org/GLM-5.3-Flash. See Available Models for model IDs and capabilities, or list models through the API.

cURL
Python

Run this in a macOS or Linux shell. No SDK installation is required:

curl --request POST \
  --url "https://api.telnyx.com/v2/ai/openai/chat/completions" \
  --header "Authorization: Bearer $TELNYX_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
    "model": "zai-org/GLM-5.3-Flash",
    "messages": [
      {"role": "user", "content": "Describe Telnyx in one sentence."}
    ]
  }'

A successful request returns JSON with the answer in choices[0].message.content. The usage object reports token counts. The answer text varies between requests.

Create a virtual environment and install the OpenAI SDK:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade openai

py -m venv .venv
.venv\Scripts\Activate.ps1
python -m pip install --upgrade openai

Save this as quickstart.py:

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["TELNYX_API_KEY"],
    base_url="https://api.telnyx.com/v2/ai/openai",
)

response = client.chat.completions.create(
    model="zai-org/GLM-5.3-Flash",
    messages=[
        {"role": "user", "content": "Describe Telnyx in one sentence."}
    ],
)

print(response.choices[0].message.content)

Run it in the activated environment:

python quickstart.py

The script prints the model’s final answer. A reasoning model may take time to produce it; use streaming below to see output as it arrives.

The SDK base URL ends in /openai; the full HTTP endpoint ends in /openai/chat/completions. See the Chat Completions API reference for all request parameters and response fields.

3. Stream the response (optional)

In quickstart.py, keep the imports and client setup, then replace the request and final print with this code. Run python quickstart.py again:

stream = client.chat.completions.create(
    model="zai-org/GLM-5.3-Flash",
    messages=[
        {"role": "user", "content": "Describe Telnyx in one sentence."}
    ],
    stream=True,
)

reasoning_started = False
answer_started = False
for chunk in stream:
    if not chunk.choices:
        continue
    delta = chunk.choices[0].delta
    reasoning = getattr(delta, "reasoning_content", None)
    if reasoning:
        if not reasoning_started:
            print("--- reasoning ---")
            reasoning_started = True
        print(reasoning, end="", flush=True)
    if delta.content:
        if not answer_started:
            print("\n--- answer ---")
            answer_started = True
        print(delta.content, end="", flush=True)
print()

Reasoning models can return reasoning_content separately from the final answer in content: on message for a non-streaming response, or delta when streaming. Models without reasoning omit this field. The example prints both separately so the reasoning phase does not appear silent. Omit the reasoning print block to display only the answer in an application.

The examples use the model’s default reasoning effort. Set reasoning_effort only to a value supported by the selected model; unsupported values return an error.

Troubleshooting

Symptom	What to check
`KeyError: 'TELNYX_API_KEY'`	Set the environment variable in the terminal running Python, then rerun the script.
Authentication fails (`401`)	Use an active Telnyx API v2 key and the Telnyx base URL. Check whether the key has expired or been revoked.
The request rejects a model or parameter	Copy the exact model ID from Available Models, and check supported parameters in the API reference.
Rate limit reached (`429`)	Reduce request volume and retry with backoff. See Inference rate limits.
No answer appears yet	A reasoning model can emit `reasoning_content` before `content`. Use the streaming example to see progress. If `finish_reason` is `length`, increase `max_tokens`; this budget includes both reasoning and answer tokens.

Continue the conversation

Send prior turns in messages to preserve context. Use system for model instructions, user for input, assistant for previous responses, and tool for function results. See Function Calling to connect the model to application code.

What Next?

I want to…	Go to
Build a voice assistant	Voice Assistant Quickstart
Call custom code from the model	Function Calling / Streaming Functions
Ground responses in documents	Embeddings
Identify themes in data	Clusters
Browse all models	Available Models
Check costs and limits	Inference Pricing / Spending Limits

Inference

Data Residency

Integrations

Tutorials

API Reference

For AI Agents

Inference API Quickstart

Prerequisites

1. Set your API key

2. Send a chat completion

3. Stream the response (optional)

Troubleshooting

Continue the conversation

What Next?

​Prerequisites

​1. Set your API key

​2. Send a chat completion

​3. Stream the response (optional)

​Troubleshooting

​Continue the conversation

​What Next?

Prerequisites

1. Set your API key

2. Send a chat completion

3. Stream the response (optional)

Troubleshooting

Continue the conversation

What Next?