Based on OpenAI’s official docs, checked October 27, 2026

GPT-6.1 Terra in Codex and the API: setup, pricing and limits

GPT-6.1 Terra is OpenAI’s fast, lower-cost model in the GPT-6.1 family. OpenAI makes it available in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, and in the API as gpt-6.1-terra. This guide covers using it as a coding agent: selecting it in Codex, calling it from the Responses API, what it costs, its limits, and when OpenAI says a larger model is the better choice. Everything here comes from OpenAI’s announcement, model docs and Codex docs, checked October 27, 2026.

1. Update Codex CLI
OpenAI’s Codex models page says GPT-6.1 Terra needs Codex CLI 0.159.0 or later.
```
codex --version
```
2. Start Codex with Terra
Pass the model with -m, or switch inside a running session with /model. In the Codex app, choose it under Power > Advanced.
```
codex -m gpt-6.1-terra
```

3. Call it from the API

Tool calling requires the Responses API. Reasoning effort can be none, low, medium (the default), high, xhigh or max.

curl https://api.openai.com/v1/responses \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-6.1-terra","reasoning":{"effort":"low"},"input":"Summarize this diff."}'

4. Pick an effort level per task
OpenAI reports Terra’s error rate falling from 53.6% for GPT-5.6 Terra to 16.7% at low effort and 4.3% at max effort, so start low for routine edits and raise effort for harder tasks.
```
{"model":"gpt-6.1-terra","reasoning":{"effort":"max"},"input":"Find the root cause of this failing test."}
```

Pricing

Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens, which OpenAI says is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol.

Cache writes cost $1.25 per million tokens for 30-minute retention or $2.50 for 12-hour retention. Prompts over 272K tokens are billed at 2x for input and cache and 1.5x for output. Batch and Flex are 50% off, regional processing adds 10%, and fast mode costs 2x.

Context, inputs and endpoints

The context window is 1,050,000 tokens, with up to 922,000 input and 128,000 output tokens. The knowledge cutoff is June 1, 2026. It accepts text and image input.

It works with Chat Completions, Responses and Batch. Supported tools include web_search, file_search, image_generation, code_interpreter, hosted_shell, apply_patch, skills, computer_use, mcp and tool_search.

Rate limits

OpenAI’s model page lists 5,000 requests and 1M tokens per minute on the Build tier, 10,000 and 4M on Launch, and 15,000 and 40M on Grow.

What OpenAI claims about performance

On Agents’ Last Exam, OpenAI says Terra scores 15 points above GPT-5.6 Terra at about a third of the cost, and matches GPT-6 Astra at a tenth of its per-token price. It reports Terra as more than 3x faster than Astra and Sol.

On Terminal-Bench 4.0 at max effort, OpenAI reports 50.3% against 21.5% for GPT-5.6 Terra. These are OpenAI’s own figures.

When to use Sol or Astra instead

OpenAI’s announcement says Sol or Astra remain the better choice for the hardest coding work. The Codex models page describes Terra as a fast, efficient choice for writing, visual understanding and computer use.

OpenAI also says Terra meets the Critical cybersecurity threshold under its Preparedness Framework.

Tips

Use /model to drop to Terra for routine edits and switch back to a larger model for hard refactors.
Keep prompts under 272K tokens where you can; above that, input and output prices go up.
Use Batch or Flex for non-urgent jobs such as bulk code review to halve the price.