OpenAI Launches GPT-5.6 Model Family and New Work Platform Features

OpenAI recently launched the GPT-5.6 model family, consisting of three tiers named Sol, Terra, and Luna. These models are designed to serve as a price-performance ladder, with Sol acting as the flagship for complex tasks, Terra offering mid-range capabilities, and Luna providing a high-volume, cost-effective option. The release includes significant updates to the OpenAI product ecosystem, such as ChatGPT Work, a new desktop application that integrates Codex, and a beta feature called Sites for hosting AI-generated applications.

The technical focus of this release centers on agentic performance, coding efficiency, and inference orchestration. OpenAI claims that GPT-5.6 Sol sets a new benchmark for agentic and coding tasks while improving the economics of complex workflows. Independent evaluations generally place the models at the frontier of current AI capabilities, though some analysts noted persistent challenges with complex text layouts and chart transcription. Additionally, OpenAI reported that internal use of these models has significantly increased experiment throughput for its own researchers, fueling discussions about the potential for AI to automate parts of the model-training process.

This launch is significant because it signals OpenAI’s transition from a model provider to a full-stack work platform. By bundling orchestration tools, browser capabilities, and file-handling features, the company is positioning its technology to manage multi-step, long-horizon tasks. While the release has been praised for its cost efficiency and agentic power, it has also sparked debate regarding safety, specifically concerning the ease of jailbreaking the models for cyber-related tasks.