OpenAI Announces GPT-5.6 Series: Sol, Terra, and Luna Models

OpenAI has launched a limited preview of its GPT-5.6 model series on June 26, 2026, introducing three distinct models: Sol (flagship), Terra (balanced), and Luna (budget).

Pricing and Positioning

Terra, described as "a balanced model for everyday work," offers competitive performance to GPT-5.5 while being 2x cheaper, according to OpenAI. Luna is positioned as "a fast and affordable model" at the company's lowest cost point. Specific pricing per million tokens was not disclosed.

Phased Release Strategy

The models are launching through a controlled rollout process. OpenAI is starting with "a limited preview for a small group of trusted partners" before broader release. The company states it plans to make all three models generally available "in the coming weeks."

Government Coordination

As part of what OpenAI describes as "ongoing engagement with the U.S. government," the company previewed the models and their capabilities to government officials ahead of the public announcement. At government request, OpenAI is limiting initial access to trusted partners "whose participation has been shared with the government."

This represents a more structured coordination process between an AI company and U.S. authorities compared to previous model releases.

Technical Details

OpenAI has not yet released specific technical specifications including:

• Context window sizes
• Parameter counts
• Benchmark scores (MMLU, HumanEval, etc.)
• Training data cutoff dates
• Exact pricing per million tokens

The company characterizes Sol as "next-generation" and its flagship offering, suggesting it represents the most capable model in the 5.6 series.

What This Means

The three-tier model structure mirrors industry trends toward offering differentiated options for different use cases and budgets. The 2x cost reduction for Terra while maintaining competitive performance with GPT-5.5 suggests meaningful efficiency improvements in training or inference. The government preview requirement indicates evolving norms around AI safety and national security considerations for frontier models, though the practical impact of this coordination process remains unclear. The "coming weeks" timeline for general availability means developers should expect limited access in the near term.