OpenAI: GPT-6 Luna

GPT-6 Luna is a text model from OpenAI, part of the GPT-6 series. OpenAI positions it as the fast, cost-efficient member of that series, sitting below GPT-6 Sol. It was released by the provider on 22 September 2026. If you are choosing between models in the GPT-6 family, Luna is the one aimed at workloads where speed and throughput matter more than top-end capability.

On the input side, Luna accepts text, images and files, so you can send prompts that mix written instructions with visual or document content. Output is text only. The context window is 1,050,000 tokens, which is large enough to hold entire codebases, long document sets or extended conversation histories in a single request. The maximum output is 128,000 tokens, so long generations are possible alongside that large input budget.

The combination of a very large context window and mixed input modalities points to a few practical uses. Chat applications can carry long histories without aggressive truncation. Classification and extraction tasks can take whole documents, including scanned pages or screenshots, and return structured text. Lightweight agentic workflows fit as well, since the model supports tool calling and can operate over large working contexts. High-volume and latency-sensitive jobs are the target, according to OpenAI's own description, so batch processing, routing, tagging and similar repetitive work are reasonable candidates.

You call the model through api14's OpenAI-compatible API using the model id "openai/gpt-6-luna". The supported request parameters are include_reasoning, max_completion_tokens, max_tokens, reasoning, reasoning_effort, response_format, seed, structured_outputs, tool_choice, tools and verbosity. Because the endpoint follows the OpenAI chat completions shape, existing clients and SDKs work with a change of base URL and model id. Use response_format or structured_outputs when you need predictable JSON, tools and tool_choice for function calling, and seed when you want more repeatable sampling. The reasoning and reasoning_effort parameters control how much internal reasoning the model performs, while verbosity influences the length of its answers, and max_tokens or max_completion_tokens cap the generation.

Keep the limits in mind when deciding. Output is text only, so image generation or audio output is not part of this model. The provider description places Luna below GPT-6 Sol in the series, so if a task demands the strongest reasoning the family offers rather than speed and efficiency, another model may be the better fit. For everything else, particularly large-context chat, document and image understanding, classification and tool-driven agents at volume, Luna is the natural starting point.