OpenAI releases open-weight models gpt-oss-120b and gpt-oss-20b under Apache 2.0

OpenAI/gpt-oss

OpenAI released gpt-oss-120b and gpt-oss-20b, open-weight language models under Apache 2.0 licence. The 120B model nearly matches o4-mini on reasoning benchmarks while running on a single 80GB GPU; the 20B model matches o3-mini and runs on 16GB devices, enabling broad deployment.

• Two models: 120B (runs on single 80GB GPU) and 20B (runs on 16GB memory); both open-weight under Apache 2.0.
• 120B nearly matches o4-mini reasoning performance; 20B matches o3-mini capability.
• Anyone can download and run locally; designed for broad accessibility.
• Fills gap between capability and deployment simplicity for open models.