OpenAI GPT-5.6 Sol and ChatGPT Work target enterprise AI

GPT‑5.6 and ChatGPT Work together mark a clear shift in OpenAI’s strategy from providing a powerful model to owning the entire work execution stack. Sol, Terra and Luna are not just incremental upgrades; they are tuned for long‑horizon, tool‑using, multi‑agent workflows, and priced aggressively enough to pressure Anthropic’s Fable 5 and other frontier models on cost per unit of useful work rather than abstract benchmarks. That’s exactly where the AGI race is heading: from raw IQ scores to sustained, reliable agency over complex tasks.

ChatGPT Work is equally important. By wiring GPT‑5.6 directly into a first‑party agent that talks to Slack, Microsoft 365, Google Drive and other systems, OpenAI is trying to become the default orchestrator of white‑collar workflows. This raises the competitive stakes: Anthropic and others now have to match not only model quality but also the surrounding agent platform, governance tooling and pricing. The fact that US regulators forced a delay and technical review before broad release underscores how frontier models are now treated more like dual‑use infrastructure than consumer apps, and future rollouts will likely have to navigate similar scrutiny.