OpenAI Previews GPT-5.6 Sol Series and Introduces Tiered Agentic Pricing
OpenAI has launched a limited preview of its next-generation GPT-5.6 model series, introducing three distinct capability tiers: Sol (flagship), Terra (balanced everyday model), and Luna (fast, low-cost model). The release marks a major architectural transition toward long-horizon reasoning and agentic workflows, featuring a new max reasoning effort and an ultra mode that orchestrates subagents to solve complex, multi-step tasks.1
In a significant policy shift, OpenAI previewed the models' capabilities to the U.S. government ahead of launch and restricted the initial preview to a small group of trusted partners.2 The lab is coordinating with the Administration to align future releases with an upcoming Executive Order framework on cybersecurity.
Capabilities and Safeguards
GPT-5.6 Sol demonstrates a performance step-change on agentic and technical benchmarks, scoring 91.9% on Terminal-Bench 2.1 (compared to 78.9% for GPT-5.5) and exhibiting competitive exploitation capabilities on ExploitBench while using only one-third of the output tokens of previous models.
To mitigate the offensive risks of these advanced cyber capabilities, OpenAI implemented a layered safety stack, including real-time output classifiers that can pause generation for review by a larger reasoning model. Additionally, the lab spent over 700,000 A100-equivalent GPU hours on automated red-teaming to harden the models against universal jailbreaks.
"During this preview, we will continue testing and coordinating closely with partners as we work toward broader availability. We don’t believe this kind of government access process should become the long-term default... We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks, while we work with the Administration to develop the cyber Executive Order framework and a repeatable process for future model releases."
Pricing and Tier Structure
OpenAI is transitioning away from transient names toward durable capability tiers priced per 1 million tokens:
- Sol (Flagship): $5.00 input / $30.00 output
- Terra (Balanced): $2.50 input / $15.00 output
- Luna (Fast/Affordable): $1.00 input / $6.00 output
"In this new naming system introduced with GPT‑5.6, the number identifies a model’s generation, while Sol, Terra, and Luna identify durable capability tiers that can advance on their own cadence. Together, the family gives people and developers clearer choices across intelligence, speed, and cost."
-
An instance of Dynamic reasoning controls shift the cost of intelligence to adjustable inference-time compute. — This shows how Next-gen AI networks allow users to active decide on pricing tiers and active inference-time reasoning efforts. ↩︎
-
An instance of National security mandates are turning public frontier AI launches into gated, state-vetted handovers. — OpenAI's policy of pre-release previews and restricted availability showcases how state-level vetting is replacing simple commercial releases for frontier systems. ↩︎