gpt-oss Models Match Premium Power, No Cloud Needed

🎁 Ace quick missions & earn crypto rewards while gaining real-world Web3 skills. JOIN NOW! 🔥

OpenAI's gpt-oss Models Match Premium Power, No Cloud Needed

Written by Aaron S.,

Editor-In-Chief

Last Updated: August 06, 2025

Key Takeaways

OpenAI released two free AI models that run on consumer hardware with as little as 16GB RAM;
The models support up to 128,000-token inputs and perform similarly to OpenAI’s o4-mini;
Built-in safety features help the models avoid harmful responses and resist prompt attacks.

Ace quick missions & earn crypto rewards while gaining real-world Web3 skills. Participate Now! 🔥

BITDEGREE IN YOUR SOCIAL FEED

OpenAI has launched two new artificial intelligence (AI) language models that are free to use and can run directly on consumer computers.

The larger model, called gpt-oss-120b, works with one 80GB graphics card. A smaller version, gpt-oss-20b, needs 16GB of memory.

Both models support long conversations or inputs, up to 128,000 tokens, which matches the limit set by OpenAI’s GPT-4o.

What Is Chia? | Crypto Finally Explained

Did you know?

Want to get smarter & wealthier with crypto?

Subscribe - We publish new crypto explainer videos every week!

In an August 5 announcement, OpenAI stated that the new models are designed to perform well in real-world use while still being easy to run on common devices.

These models are available under the Apache 2.0 license, which allows both personal and commercial use. OpenAI said they perform about as well as its o4-mini model when tested on tasks that require reasoning.

Instead of using all parameters at once, the models use a "mixture-of-experts" design, which activates only part of the full system for each task.

Developers can choose how much processing power to use by selecting from three levels of reasoning effort: low, medium, or high. These settings are easy to adjust with a short instruction in the system message.

OpenAI trained the models using techniques from earlier systems like o3, including reinforcement learning and fine-tuning. After training, they went through an extra process to make them safer and more reliable.

The models were also trained to avoid harmful answers and resist tricks that try to bypass safety controls.

Recently, OpenAI announced plans to use a new data center in northern Norway to support its AI operations in Europe. What did the company say? Read the full story.