gpt-oss-120b, an open-weight reasoning model built by OpenAI, is now available on Telnyx Inference in beta, hosted on Telnyx-owned GPUs in US regions.
What's new
- Open-weight reasoning from OpenAI: Telnyx serves gpt-oss-120b, a 117B mixture-of-experts model, with model=openai/gpt-oss-120b in the Chat Completions API. It offers a 128K context window and low, medium, and high reasoning effort, and is not currently available in AI Assistants.
- More open-weight models from US companies: gpt-oss-120b (OpenAI) and google/gemma-4-26B-A4B-it (Google) are both open-weight releases from US companies. Gemma 4 26B A4B is a lightweight 25.8B-parameter model with a 256K context window, already available on Telnyx Inference.
Why it matters
OpenAI built gpt-oss-120b in the US and released it as open weights, and Telnyx serves it on Telnyx-owned GPUs inside the United States. Because we own the infrastructure, adding models like this is fast and pricing stays at a fraction of proprietary APIs.
Example use cases
The model documentation positions gpt-oss-120b for open-weight reasoning. With a 128K context window and configurable reasoning effort, it fits reasoning-heavy workloads like document analysis and agentic planning, and teams already running OpenAI-compatible clients can switch the base URL to Telnyx without changing their code.
Getting started
- Call the chat completions endpoint with model=openai/gpt-oss-120b:
- List available models through the model catalog API to see the hosted lineup.
