OpenAI: GPT-5.6 Luna (batch)

GPT-5.6 Luna (batch) is a text model from OpenAI, part of the GPT-5.6 series. OpenAI describes it as a fast, cost-efficient entry in that series, aimed at high-volume and latency-sensitive work. The batch variant is served through api14 under the model id openai/gpt-5.6-luna:batch, so you can route requests to it without changing your existing OpenAI-compatible client code.

The model accepts text, image and file inputs, and it returns text only. That means you can send prompts that mix written instructions with images or uploaded documents, but you should not expect audio or other output types. Its context window is 1,050,000 tokens, which is large enough to hold long documents, extended conversation histories or sizeable code and data extracts in a single request. The maximum output is 128,000 tokens, and the knowledge cutoff is 2026-02-16. Supported request parameters include max_tokens, response_format, structured_outputs, seed, tool_choice, tools, verbosity, and the reasoning controls reasoning, reasoning_effort and include_reasoning.

Those capabilities point to a few practical patterns. Because input can include images and files alongside text, the model fits document question answering, extraction from scanned or uploaded material, and summarisation of long reports or transcripts. Structured outputs and response_format make it usable for classification, tagging and other tasks where you need machine-readable results rather than free prose. Tool and function calling support suits lightweight agentic workflows, where the model decides which tool to invoke and then continues from the result. The large context window helps when you would otherwise have to chunk a long document or maintain a lengthy running conversation. The batch designation suggests workloads where you can tolerate asynchronous turnaround in exchange for efficiency on large volumes of requests.

To call it through api14, use the OpenAI-compatible chat completions endpoint and set the model field to openai/gpt-5.6-luna:batch. Requests follow the same shape as other OpenAI-compatible calls: a messages array, plus any of the supported parameters such as max_tokens, response_format, structured_outputs, seed, tools, tool_choice, verbosity, reasoning, reasoning_effort and include_reasoning. If your application already targets an OpenAI client library, you typically only need to change the base URL to api14 and swap in this model id. Live pricing for the model is shown in the price table on this page rather than here.

One thing to weigh is the batch aspect. If your application depends on immediate, synchronous responses per request, a non-batch variant of the same model may be a better fit, since batch serving is generally oriented toward jobs that can be submitted and collected later. If you need output in a form other than text, this model will not serve that need. Within those bounds, GPT-5.6 Luna (batch) is a reasonable choice for high-volume text work that also involves images or files, structured outputs, tool use and long context.