Anthropic and OpenAI Launch Low-Cost Models in Pricing War
Anthropic and OpenAI released more affordable AI models on September 22, slashing prices to attract business customers and compete with open-weight models from China.
On September 22, 2026, Anthropic launched Claude Opus 5.5, a model designed for coding and knowledge work that matches the performance of its flagship Fable 5.1 but costs 40% less to operate than Opus 5. The model is 30% faster and features API price cuts of 20% for tokens and 60% for cache reads. Anthropic emphasized safety, noting the model is 85% less likely to bypass containment boundaries in internal tests and includes strict safeguards for biology and cybersecurity research.
Approximately 90 minutes later, OpenAI responded by releasing GPT-6 Sol and GPT-6 Luna. These models, positioned as everyday work versions of the top-tier GPT-6 Astra, feature API costs 50% lower than GPT-5.6 promotional rates. OpenAI claims GPT-6 Sol reduces factual errors by 50% and outperforms Anthropic's models in several areas, though these claims rely on internal benchmarks.
This aggressive pricing competition follows a public agreement between Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to pace the frontier of AI development to ensure safety keeps pace with capability. The shift toward high-efficiency, lower-cost models reflects pressure to generate returns on investment and competition from efficient open-weight models from China. The launch occurs as Anthropic prepares for an initial public offering with a potential valuation approaching $2 trillion, while OpenAI has delayed its IPO until next year.