OpenAI: GPT-5.6 Terra (batch)

GPT-5.6 Terra is a text model from OpenAI, part of the GPT-5.6 series. Within that series it sits in the middle: OpenAI describes it as a balanced model positioned between the flagship Sol tier and the cost-efficient Luna tier. The batch variant is the version exposed here, and the model id on api14 is openai/gpt-5.6-terra:batch. Its knowledge cutoff is 2026-02-16.

Terra accepts text, images and files as input, and it returns 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, and the maximum output is 128,000 tokens. Because it takes images and files as well as text, it can work over screenshots, scanned pages, diagrams and uploaded documents alongside ordinary prompts. It also supports reasoning: the reasoning, reasoning_effort and include_reasoning parameters let you control how much internal deliberation the model does and whether that reasoning is surfaced back to you.

Those capabilities point to a fairly wide set of practical uses. Everyday coding is the first: writing, reviewing and refactoring code, with whole repositories fitting inside the context window. Reasoning tasks over long inputs follow from the same combination, such as analysing lengthy contracts, research papers or log files, or answering questions that require pulling details from several documents at once. The image and file inputs make it usable for extracting structured data from screenshots, forms and PDFs. Tool calling and structured outputs make it a reasonable fit for agentic workflows, where the model decides which tools to invoke and returns results in a schema your code can parse.

You call it through api14's OpenAI-compatible API using the model id openai/gpt-5.6-terra:batch. The supported request parameters are include_reasoning, max_tokens, reasoning, reasoning_effort, response_format, seed, structured_outputs, tool_choice, tools and verbosity. In practice that means you can cap output length with max_tokens, tune reasoning depth with reasoning_effort, request JSON or schema-conforming replies with response_format and structured_outputs, define callable functions with tools and tool_choice, and use seed when you want reproducible sampling. Because the endpoint is OpenAI-compatible, existing OpenAI client libraries work with only the base URL and model id changed.

One thing to keep in mind is that Terra outputs text only, so it is not the right choice if you need generated images, audio or video. The batch variant is also worth weighing against the standard endpoint if your workload is interactive, since batch processing is generally aimed at work that can be submitted and collected rather than answered immediately. If you need OpenAI's top tier rather than a balanced one, the Sol tier is the alternative named in the series; if cost efficiency matters more than capability, Luna is the other.