OpenAI's API offers various models, including GPT-6 Astra and GPT-6 Luna, each suited for different applications. Understanding their described functionalities can help developers choose the appropriate model for their needs.
GPT-6 Astra vs Luna UK AI code review: evidence and context
GPT-6 Astra is designed for complex reasoning, coding, computer use, research, and document creation. Its capabilities extend to tasks requiring intricate thought processes and detailed output. The model supports various levels of reasoning effort, including low, medium, high, xhigh, and max. Pricing for prompts exceeding 272K input tokens is adjusted, with input and cache rates at 2x and output rates at 1.5x for the full request. Specific pricing tiers like Batch and Flex are set at 50% of Standard rates, while Fast mode operates at 2x the applicable rates.
Source: https://developers.openai.com/api/docs/models/gpt-6-astra
GPT-6 Luna for High-Volume Tasks
GPT-6 Luna is presented as the most efficient model for focused, high-volume tasks. It supports a range of reasoning efforts, from none to max, with medium being the default. Function calling is available through the Responses API, and with Chat Completions, it is supported only when reasoning_effort is set to none. EU data residency is an option for Standard, Fast, Flex, and Batch processing. Pricing for cached input tokens is set at 10% of the uncached input token rate.
Source: https://developers.openai.com/api/docs/models/gpt-6-luna
Model Selection Considerations
When selecting between models, developers can refer to the API model catalog for available versions and pricing. Luna is identified as the most cost-efficient model, whereas Astra is described as state-of-the-art. Guidance is available to assist in choosing a model to reduce costs or latency.
Source: https://developers.openai.com/api/docs/guides/model-selection