OpenAI models Archives

Winsage

August 6, 2025

Microsoft makes OpenAI’s new open model available on Windows

OpenAI has released a new, free, and open GPT model called gpt-oss-20b, which can run on personal computers. Microsoft is facilitating its integration for Windows users through the Windows AI Foundry, with plans to extend support to macOS. The model requires a PC or laptop with at least 16GB of VRAM and is optimized for code execution and tool utilization. Microsoft has pre-optimized the model for local inference, indicating potential future support for more devices. This is the first instance of an OpenAI model running locally on Windows, coinciding with Amazon's adoption of the new open-weight GPT-OSS models for its cloud services.

Winsage

August 6, 2025

OpenAI’s New Open Models Accelerated Locally on NVIDIA GeForce RTX and RTX PRO GPUs

NVIDIA has partnered with OpenAI to enhance the gpt-oss models for NVIDIA GPUs, enabling rapid inference and supporting millions of users on NVIDIA RTX AI PCs. The gpt-oss-20b and gpt-oss-120b models, trained on NVIDIA H100 GPUs, feature open-weight reasoning and can handle context lengths of up to 131,072 tokens. Users can utilize these models through frameworks like Ollama, which provides a user-friendly interface for experimentation. The models are optimized for RTX GPUs and support applications such as web search and coding assistance. Developers can also access the models via Microsoft AI Foundry Local and other frameworks, with NVIDIA contributing to open-source projects to enhance performance.

Winsage

August 6, 2025

OpenAI open weight models now available on AWS

AWS has introduced two new OpenAI models with open weights, the gpt-oss-120b and gpt-oss-20b, available through Amazon Bedrock and Amazon SageMaker JumpStart. These models are designed for text generation and reasoning tasks, excelling in coding, scientific analysis, and mathematical reasoning, with performance comparable to leading alternatives. They support a context window of 128K and adjustable reasoning levels (low, medium, high). Users can access these models via an OpenAI-compatible endpoint in Bedrock, utilizing the OpenAI SDK or Bedrock APIs for integration. To access the models in Amazon Bedrock, users must request access through the console and can evaluate them using the Chat/Test playground. The process for using the OpenAI SDK involves configuring the API endpoint and authentication with an Amazon Bedrock API key. Users can build AI agents using frameworks that support the Amazon Bedrock API or OpenAI API, with deployment facilitated by Amazon Bedrock AgentCore. The OpenAI models are available in the US West (Oregon) region through Amazon Bedrock and in the US East (Ohio, N. Virginia) and Asia Pacific (Mumbai, Tokyo) regions via SageMaker JumpStart. Each model provides full chain-of-thought output capabilities for enhanced interpretability. They can be modified and customized to meet specific needs, and security measures are in place to ensure robust evaluation processes. Compatibility with the standard GPT-4 tokenizer is also included. For cost details, users can refer to the Amazon Bedrock and SageMaker AI pricing pages.