{
  "id": 9796294,
  "title": "GPT-6 Sol and Luna: What OpenAI’s Cheaper Models Mean for AI Agents",
  "url": "https://urgent.news/2026/09/25/gpt-6-sol-and-luna-what-openais-cheaper-models-mean-for-ai-agents",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-25T11:14:06.000Z",
  "source": {
    "name": "TechRepublic",
    "slug": "techrepublic",
    "url": "https://www.techrepublic.com/article/news-openai-gpt-6-sol-luna-api-costs/"
  },
  "original_language": "en",
  "account": "OpenAI has reduced the prices of its latest AI models, GPT-6 Sol and Luna, by 50%, making them more affordable for developers building AI agents. GPT-6 Sol, designed for complex tasks like coding and long-running agents, costs $2 per 1 million input tokens and $10 per 1 million output tokens. Luna, intended for high-volume jobs such as summarization, is priced at $0.10 and $0.50 respectively. This price cut could enable AI agents to run for longer periods without the cost becoming a limiting factor. In benchmark tests, GPT-6 Sol performed better than its predecessor, GPT-6 Astra, with a 33.2% score at maximum reasoning effort compared to 30.3% for GPT-6 Astra. Similarly, GPT-6 Luna improved upon its predecessor, GPT-5.6 Luna, by increasing the AutomationBench score by 5.4 percentage points at high reasoning effort while cutting the cost per task by 58%. These lower prices are expected to have a significant economic impact on AI agent development, allowing developers to run more tasks within their existing budgets. The shift towards matching model cost to task difficulty rather than defaulting to the most capable model may become a key trend in the AI landscape. However, it's important to note that OpenAI's claims are based on their own evaluations, and real-world performance may vary.",
  "summary": "OpenAI launched GPT-6 Sol and Luna with lower API prices, cheaper caching and stronger coding results aimed at reducing the cost of AI agents. The post GPT-6 Sol and Luna: What OpenAI’s Cheaper Models Mean for AI Agents appeared first on TechRepublic .",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}