OpenAI Launches GPT-5.6 Model Series With Three Tiers, New Safety Features, and Competitive Pricing

OpenAI launches the GPT-5.6 model series in limited preview, offering three tiers—Sol, Terra, and Luna—with competitive pricing, record-breaking inference speeds up to 750 tokens per second via Cerebras, and the company's most advanced safety stack yet, featuring over 700,000 GPU hours of automated red-teaming.

• OpenAI is launching a limited preview of the GPT-5.6 model series, featuring three tiers: Sol (flagship), Terra (balanced, 2x cheaper than GPT-5.5), and Luna (fast and affordable), with broader availability planned in the coming weeks.
• GPT-5.6 Sol debuts with OpenAI's most robust safety stack to date, featuring layered safeguards including real-time misuse classifiers, account-level monitoring, and over 700,000 A100-equivalent GPU hours dedicated to automated red-teaming to defend against jailbreaks and adversarial attacks.
• The GPT-5.6 series is priced per 1M tokens with Sol at $5 input/$30 output, Terra at $2.50 input/$15 output, and Luna at $1 input/$6 output, and OpenAI is also launching Sol on Cerebras at up to 750 tokens per second in July for unprecedented inference speed.