The documented model ID is openai/gpt-6.1-terra; a latency-oriented variant is available as openai/gpt-6.1-terra-fast.
According to Vercel, the model accepts text and images, returns text, and has a context window of more than one million tokens. It supports configurable reasoning effort, structured outputs, tool calling, and implicit and explicit prompt caching. Vercel describes it as intended for visual reasoning, computer use, and writing, including document and desktop-app workflows.
The model can be called through Vercel's AI SDK, its OpenAI-compatible Chat Completions API, the Responses API, and coding agents connected to AI Gateway. Vercel says Codex and Cursor users can configure the gateway via the latest Vercel CLI. The company also says AI Gateway passes through provider pricing without markup and charges no inference platform fee, including for BYOK requests.
The announcement does not provide model prices, benchmark results, availability by region, or independent measurements of latency, factual accuracy, or coding performance.