OpenAI: GPT-5.2 Chat

GPT-5.2 Chat is a text model from OpenAI, released by the provider on 10 December 2025. It is described as the fast, lightweight member of the 5.2 family, optimized for low latency chat while retaining strong general intelligence, and it uses adaptive reasoning to selectively think when a task calls for it. On api14 it is listed under the main category of text, and the model id you use when calling it is openai/gpt-5.2-chat.

The model accepts text, images and files as input, and it returns text. Its context window is 128,000 tokens, with a maximum output of 32,000 tokens, so it can hold long conversations, large documents or a mix of written and visual material in a single request while still producing substantial responses. Because input is multimodal, you can pass screenshots, scanned pages or attached files alongside your prompt rather than converting everything to plain text first. Output is text only, so any image or file you send is there to inform the answer, not to be transformed into another media type.

Those capabilities point to a fairly wide set of practical uses. Low latency chat makes it a reasonable fit for interactive assistants, customer support flows and other applications where a user is waiting on each reply. The 128,000 token context supports summarization, question answering and analysis over long documents, contracts or codebases, and the file and image inputs help with tasks such as extracting information from PDFs, reading charts or screenshots, and reviewing material that mixes prose with visuals. Tool calling and structured outputs, covered below, extend this to workflows where the model needs to fetch data or return machine readable results rather than free text.

You call the model through api14's OpenAI-compatible API using the model id openai/gpt-5.2-chat, which means existing OpenAI SDKs and client libraries work with only a base URL and key change. The supported request parameters are max_completion_tokens, response_format, seed, structured_outputs, tool_choice, tools and verbosity. In practice, tools and tool_choice let the model call functions you define, structured_outputs and response_format let you constrain the shape of the reply for downstream parsing, seed helps with reproducibility, max_completion_tokens caps the generated length, and verbosity adjusts how much the model says. Note that temperature and top_p are not in the supported list, so control over sampling is limited to what is listed.

It is worth checking the live price table on this page before committing, since that is where current rates are shown. If your workload depends on parameters outside the supported set, or if you need output modalities other than text, another model may fit better. For straightforward, latency sensitive chat and document tasks within a 128,000 token context, though, this model covers the common cases without extra configuration.