Daily AI CatchupOpenAI Moves GPT-5.6 Into Preview With Three Capability Tiers
OpenAI has moved GPT-5.6 into a narrow preview with three tiers — Sol, Terra, and Luna — including a new reasoning-effort control slider and an "ultra" mode for complex tasks. Broad access depends on US government review approval, a new dynamic in the AI release pipeline reflecting how quickly regulation is entangling frontier deployments.
Why it matters
💻 Developer · A reasoning-effort slider across three tiers is worth planning API integration around now, ahead of general availability, so you're not scrambling when it ships widely.
📦 Product · Government review as a gating step for broad model access is now a real planning variable — build buffer time into any roadmap that depends on next-generation OpenAI models.
🎨 Design · No direct design impact yet — wait for general availability before committing UI work to tier-specific behavior.
📈 Business · Regulatory review before broad release is becoming a standard part of frontier model rollouts across labs, not just Anthropic's export-control saga — worth normalizing in vendor-risk planning.
🤔 Just Curious · OpenAI's next major AI model is coming in three versions of different power levels, but it needs government approval before everyone can use it — a new kind of hurdle for releasing powerful AI.
Sources: OpenAI might be preparing GPT-5.6 for next week's release