GPT-OSS 120B

GPT-OSS 120B is OpenAI's first major open-source model, marking a significant shift in the company's strategy. At 120 billion parameters, it is a dense Transformer model that delivers strong general-purpose performance across coding, reasoning, and knowledge tasks — making it a compelling alternative to proprietary models for teams that need self-hosting capability or want to avoid vendor lock-in.

The model is available on the CallMissed gateway, enabling low-latency inference globally. Its 128K context window handles substantial documents and codebases, and its performance on standard benchmarks is competitive with many proprietary models at similar price points. The open-source nature means it can be fine-tuned, quantized, and deployed on custom infrastructure.

GPT-OSS 120B serves as a strong baseline for the open-source LLM ecosystem, and several derivative models (like NVIDIA's Nemotron 3 Super) have been built on top of it. For teams that need a capable, self-hostable model with the OpenAI training methodology, GPT-OSS 120B is the go-to choice.