OpenAI launches more token-efficient GPT-5.6 AI Model Family consisting of Sol, Terra and Luna

OpenAI has announced its new GPT-5.6 artificial intelligence model family consisting of their flagship Sol, balanced Terra tier and cost-efficient Luna models. According to OpenAI they offer plenty of cost reductions, stating that the smaller Terra and Luna models match or outperform competing frontier models like Claude Fable 5 at around one-sixteenth the cost.

Improved token efficiency

The standout feature of this release is a focus on computing efficiency and autonomous multi agent workflows. GPT-5.6 Sol sets a new high score of 53.6 on the Agents Last Exam benchmark, beating Claude Fable 5 by 13.1 points while operating at lower estimated operational costs. The models are also smarter, and have enhanced judgement, while maintaining solid guardrails as well.

If ChatGPT-5.6 becomes more energy efficient and affordable to use, would you choose it over Anthropic or Gemini? We’re hopeful that it could lead to more affordable and energy efficient AI from other competitors too, but do feel free to share your thoughts in the comments below. Stay tuned to TechNave.com for more details.