Telnyx provides access to a variety of open-source and proprietary large language models (LLMs) through the Inference API. All models are accessible via the Chat Completions API and are OpenAI-compatible.
Chat Completions models
These models are verified for use with the Chat Completions API.
Currently available in the US only. Additional regions coming soon.
| Model ID | Description |
|---|
moonshotai/Kimi-K2.5 | Balance of intelligence and cost, real-time voice AI (Recommended) |
zai-org/GLM-5 | Highest intelligence open-source LLM |
MiniMaxAI/MiniMax-M2.5 | High intelligence at lower cost |
Other models
The following models are also available on the Telnyx platform.
| Model ID | Description |
|---|
Qwen/Qwen3-235B-A22B | Strong function calling and reasoning |
Groq/kimi-k2-instruct | Open-source instruction-tuned model |
Groq/llama-4-maverick-17b-128e-instruct | Open-source instruction-tuned model |
anthropic/claude-haiku-4-5 | Anthropic proprietary model |
google/gemini-2.5-flash | Google proprietary model |
openai/gpt-4.1 | OpenAI proprietary model |
openai/gpt-4o | OpenAI proprietary model |
openai/gpt-5 | OpenAI proprietary model |
openai/gpt-5.1 | OpenAI proprietary model |
openai/gpt-5.2 | OpenAI proprietary model |
fixie-ai/ultravox-v0_4_1-llama-3_1-8b | Audio language model for voice input |
Choosing a model
- General use:
moonshotai/Kimi-K2.5 offers the best balance of intelligence and cost.
- Highest quality:
zai-org/GLM-5 for tasks requiring maximum intelligence.
- Cost-sensitive:
MiniMaxAI/MiniMax-M2.5 provides high intelligence at lower cost.
- Function calling:
Qwen/Qwen3-235B-A22B is recommended for function calling workflows.
- Audio input:
fixie-ai/ultravox-v0_4_1-llama-3_1-8b for audio language models.
Using a model
Pass the model ID in the model parameter of your API request:
import os
from openai import OpenAI
client = OpenAI(
api_key=os.getenv("TELNYX_API_KEY"),
base_url="https://api.telnyx.com/v2/ai",
)
chat_completion = client.chat.completions.create(
model="moonshotai/Kimi-K2.5",
messages=[{"role": "user", "content": "Hello!"}],
)