Anthropic and OpenAI have launched cheaper versions of their artificial intelligence models, with OpenAI introducing GPT-6 Sol and GPT-6 Luna, and Anthropic unveiling Claude Opus 5.5. These releases represent the first new models from either lab since concerns about AI's potential societal threats emerged, and they aim to address customer demand for more cost-effective AI solutions.
OpenAI's new GPT-6 Sol and GPT-6 Luna models are priced 50% lower than the promotional pricing for GPT-5.6. Sol is designed for complex tasks like coding, while Luna is geared towards high-volume applications such as information extraction and document summarization. Anthropic's Claude Opus 5.5 is described as a more token-efficient version, costing approximately 40% less to operate than its predecessor, Opus 5, while maintaining performance levels comparable to its flagship Fable 5.1 model.
These price reductions come as both companies face increasing pressure from competitors offering cheaper, open-weight models, particularly from Chinese firms like Alibaba, Moonshot AI, and DeepSeek. This intensified competition is driving a price war within the AI industry, impacting the ability of OpenAI and Anthropic to profit and grow their model prices, despite their upcoming initial public offerings. Analysts note that while this trend puts pressure on margins, it is positive for the overall AI boom as lower costs are increasing usage and overall AI spending.
The releases highlight a shift in focus within the AI industry towards cost-effectiveness and accessibility, rather than solely on groundbreaking new capabilities. Both companies are passing on efficiency savings from improvements in caching and inference directly to customers. This strategy is crucial as businesses are scrutinizing their AI spending more closely, looking for optimized costs and clearer returns on investment amid an ongoing price war.
