GPT-6.1 Terra: Vercel AI Gateway Integration

One-sentence takeaway

In Vercel AI Gateway, GPT-6.1 Terra's model name is openai/gpt-6.1-terra. You can call it through the Vercel AI SDK, the OpenAI-compatible Chat Completions API, the Responses API, or a coding agent connected to Gateway. Gateway provides centralized usage and cost tracking, retries, failover, and routing configuration.

Use cases

• Suitable tasks: Connecting GPT-6.1 Terra to Vercel AI Gateway; sending requests through the AI SDK, an OpenAI-compatible API, or a coding agent; centrally managing model usage and routing.

• Unsuitable tasks: Inferring all SDK parameters, the Responses API request format, or specific routing rules from this release note; it does not provide complete configurations for these.

• Supported model versions: GPT-6.1 Terra. The standard model name is openai/gpt-6.1-terra; for latency-sensitive tasks, you can choose openai/gpt-6.1-terra-fast.

• Supported clients, agents, or APIs: Vercel AI SDK, OpenAI-compatible Chat Completions API, Responses API, and coding agents connected to AI Gateway, such as Codex and Cursor. The source also provides setup steps for fx.

• Recommended reasoning effort and parameters: The source says reasoning effort can be configured, but does not list available levels or recommended values. The API examples below use streaming; their sample prompt demonstrates drafting a proposal.

Ready-to-use content

Vercel AI SDK

The following code is preserved from the original API example:

import { streamText } from 'ai';

const result = streamText({
  model: 'openai/gpt-6.1-terra',
  prompt: 'Draft a one-page proposal for migrating our docs site to Next.js.',
});

OpenAI-compatible Chat Completions API

The original example uses the Python OpenAI SDK, points base_url to AI Gateway, and reads the Gateway API key from an environment variable:

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AI_GATEWAY_API_KEY"],
    base_url="https://ai-gateway.vercel.sh/v1",
)

stream = client.chat.completions.create(
    model="openai/gpt-6.1-terra",
    messages=[{"role": "user", "content": "Draft a one-page proposal for migrating our docs site to Next.js."}],
    stream=True,
)

for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="", flush=True)

Connecting Codex

The source gives two options: run the Vercel CLI to detect the agent automatically, create a Gateway API key, and write it to the configuration; then run /model in Codex to select GPT-6.1 Terra. Or launch Codex directly with this model:

npx vercel@latest ai-gateway setup
codex --model openai/gpt-6.1-terra

Connecting fx

The connection steps and model selection command listed in the source are:

curl -fsSL https://fx.sh/setup.sh | bash
fx

In fx, run /setup and choose either Vercel sign-in or pasting an AI Gateway key. Then run:

/model openai/gpt-6.1-terra

Testing or workflow steps

1. Choose an integration method: AI SDK, OpenAI-compatible Chat Completions, Responses API, or a coding agent. Prepare AI Gateway credentials according to the relevant client documentation; the source explicitly uses the AI_GATEWAY_API_KEY environment variable only in its Chat Completions code.

2. Set the model name to openai/gpt-6.1-terra. If the task prioritizes latency, evaluate openai/gpt-6.1-terra-fast.

3. Validate model output with representative requests, then check usage and cost in Gateway. The original release page provides no benchmark results or guarantees for quality or latency.

4. To manage availability or performance centrally, configure capabilities such as retries, failover, and routing rules in Gateway as needed. Refer to the AI Gateway documentation and account configuration for specific options.

flowchart LR
  A[Choose an SDK, API, or agent] --> B[Configure AI Gateway credentials]
  B --> C[Select openai/gpt-6.1-terra]
  C --> D[Send a representative request]
  D --> E[Check output, usage, and cost]
  E --> F[Configure retries, failover, and routing as needed]

Original evidence and data

• The release page gives openai/gpt-6.1-terra as the standard model identifier and openai/gpt-6.1-terra-fast as the fast-mode identifier.

• The release page lists the AI SDK, OpenAI-compatible Chat Completions API, Responses API, and coding agents connected to AI Gateway as integration methods. It shows JavaScript examples for the AI SDK and a Python example for Chat Completions.

• The Codex example recommends first running npx vercel@latest ai-gateway setup. After that, you can run /model to select GPT-6.1 Terra or launch Codex with codex --model openai/gpt-6.1-terra.

• Gateway features listed include calling models through a unified interface, tracking usage and cost, and configuring retries, failover, performance optimizations, and routing rules. The page also mentions custom reporting, Zero Data Retention support, and API key budgets.

• The page says Gateway charges provider prices without a markup and does not charge platform fees for inference. This also applies to BYOK requests.

• The model is described as accepting text and image input and producing text output, with a context window of more than 1M tokens. It supports configurable reasoning effort, structured outputs, tool calling, and implicit and explicit prompt caching.

Scope and limitations

• This is Vercel's official release note on AI Gateway integration methods and model identifiers. It is not an independent evaluation of GPT-6.1 Terra and provides no task-level quality, latency, or cost data.

• The English prompt in the original example is an input Vercel uses to demonstrate a call; it is not a GPT-6.1 Terra-specific prompt or a complete prompt template.

• The page only summarizes Responses API availability and does not provide request code. It also does not list available reasoning-effort values.

• The release page only summarizes Gateway's retry, failover, and routing capabilities; it includes neither detailed rule syntax nor configuration examples. AI Gateway model names and URLs are part of the Vercel integration path and should not be substituted for native OpenAI API configuration.

• The release page is dated 2026-10-26. Pricing, features, and the model catalog may change, so check current Vercel documentation before use.

Reproduction notes

Open the original page in the Tabbit browser and verify the publication date, model identifiers, integration methods, AI SDK and Chat Completions code examples, Codex/fx setup steps, and Gateway feature descriptions. The page body and examples are fully readable. The collection date was 2026-10-27.