gpt-oss playground

gpt-oss is a family of open-weight language models from OpenAI, built upon the Mixture-of-Experts (MoE) architecture. This playground allows you to experiment with the gpt-oss-120b and gpt-oss-20b models, explore their capabilities.

Model

120b
gpt-oss-120b
Designed for large-scale infrastructure
20b
gpt-oss-20b
Optimized for running on-device

Reasoning level

Token Usage

Input
0
Output
0
Total
0

I'm gptoss—how can I help?