OpenAI has launched GPT-6 Sol and Luna, reducing API pricing by 50% and enhancing performance metrics.
OpenAI Introduces New Models
OpenAI has unveiled two new AI models, GPT-6 Sol and GPT-6 Luna, significantly cutting their API pricing by 50% from the promotional rates of GPT-5.6. This launch comes shortly after Anthropic's release of Claude Opus 5.5.
Model Performance and Pricing Details
GPT-6 Sol is priced at $2 per million input tokens and $10 for output tokens, while Luna's costs are set at $0.10 and $0.50, respectively. These price reductions mark a strategic move to offer more competitive options against Anthropic's products, providing better value at every tier.
Both models are designed as faster and more affordable alternatives to OpenAI's flagship model, GPT-6 Astra, which was introduced earlier. Sol and Luna aim to cater to routine tasks without compromising much on efficiency. OpenAI's pricing strategy aims to make automated business tasks more accessible to users.
Benchmark Performance
Performance evaluations for the models are based on various benchmarks. In a recent test, GPT-6 Sol achieved a score of 33.2% for its reasoning capabilities, outperforming Claude Opus 5, which scored 26.9% at a significantly higher cost per task. Moreover, in the Agents' Last Exam, GPT-6 Sol scored 56.4%, which is 60% more cost-effective than its competitor in terms of task execution.
OpenAI has also introduced a feature to enhance reasoning effort, allowing users to allocate more resources for improved accuracy. This update is expected to further optimize the performance of the new models in handling a wide array of business workflows.
Availability and Rollout
Currently, both GPT-6 Sol and Luna are available through ChatGPT Work and Codex for various subscription tiers, including Plus, Pro, Business, Enterprise, and Edu. Additionally, Luna is accessible to free users via the desktop app. OpenAI is gradually integrating these models into its services to ensure stability as they roll out.
