OpenAI has introduced two new open-weight generative AI models, gpt-oss-120b and gpt-oss-20b. These models allow users to run configurations on their own hardware, providing a range of settings for simulated reasoning and tool use. The gpt-oss-20b model has 21 billion parameters with a mixture-of-experts technique, while the gpt-oss-120b model has 117 billion parameters requiring 80GB of memory. Both models offer robust performance and a context window of 128,000 tokens, with the smaller model capable of operating on consumer-grade machines with at least 16GB of memory.
OpenAI has announced the release of two new generative models, gpt-oss-120b and gpt-oss-20b, which are open-weight models available for personal hardware usage.
The gpt-oss-20b model features 21 billion parameters and utilizes a mixture-of-experts approach, optimizing its efficiency for consumer-level machines.
gpt-oss-120b, with its 117 billion parameters, requires significant hardware with 80GB of memory for optimal performance, fitting on specialized AI GPUs.
Collection
[
|
...
]