OpenAI has launched the GPT Work series, powered by the latest GPT-5.6 model, marking a significant advancement in AI capabilities. The new model is available in three performance tiers—Sol, Terra, and Luna—each designed to offer varying levels of performance and cost efficiency.
Benchmark Dominance
GPT-5.6 Sol, the top-tier variant, has outperformed Anthropic's Fable 5 series in key benchmarks. According to OpenAI, Sol achieved a score of 53.6 on the ALE benchmark, surpassing Fable 5 by 13.1 points. In the Artificial Analysis Coding Agent Index, Sol scored 80.0, leading Fable 5 by 2.8 points. These results highlight the model's superior reasoning and coding abilities.
Unified Desktop Experience
Alongside the model launch, OpenAI introduced a new ChatGPT desktop app that integrates Chat, Work, and Codex functionalities. This unified platform aims to streamline user interactions across diverse tasks, from casual conversation to complex coding and professional work.
Global Rollout
The GPT-5.6 model is being rolled out globally in phases, with full availability expected by the weekend. OpenAI confirmed that users across all tiers will gain access gradually, ensuring a smooth deployment.
