GPT 5.6 Terra

GPT-5.6 Terra: Model Introduction & Practical Guide

GPT-5.6 Terra is the "earth" of OpenAI's astronomy-named family - the balanced tier that most users and teams will actually run every day. Its pitch is disarmingly simple: GPT-5.5-class performance at half the price, wrapped in the same 1.5M-token context window and tooling as its flagship sibling.

Here's the short version: Terra exists because most work doesn't need Sol's $5/$30-per-million-token reasoning depth. Spreadsheet analysis, drafting, code assistance, research summarization, and agentic office flows are the everyday workload, and Terra handles them at half the cost without falling out of the frontier generation. It's the default for free and Go users on ChatGPT Work and Codex, and it's selectable across ChatGPT tiers with effort controls. The trade-offs: it is deliberately not the maximum-effort tier - for the hardest terminal tasks, genomics pipelines, or security research, Sol (or competitors' flagships) remains the right call - and it inherits the family's layered safety stack.

This guide covers model overview, core features, technical specifications, capability comparison, core advantages, recommended use cases, example prompts, and selection recommendations.

---

Quick Facts

Table of Contents

1. Model Overview
2. Core Features
3. Technical Specifications
4. Capability Comparison
5. Core Advantages
6. Recommended Use Cases
7. Example Prompts
8. Selection Recommendations
9. FAQ
10. Sources & Further Reading

---

1. Model Overview

OpenAI's three-tier GPT-5.6 lineup maps the model family to economics rather than to capability alone: Sol for the hardest tasks, Terra for the broad middle, Luna for high-volume low-cost work. Terra's brief is the largest slice of real usage - the analysis, drafting, and assistance that fills a workday - and its promise is that you don't pay flagship rates for it.

"GPT-5.5-class at half price" is a specific and checkable claim: tasks that previously required the previous generation's flagship can now run on Terra at reduced cost per token, with the same 1.5M-token context window for long documents and repositories. For organizations planning AI budgets, that arithmetic - half the price for equivalent quality - is usually the difference between piloting and deploying.

Terra's distribution is deliberately broad. It is the default model for free and Go users in ChatGPT Work and Codex, and available with effort controls across ChatGPT tiers; the API exposes it alongside Sol and Luna. That makes Terra the likely "first model" most developers meet in OpenAI's 2026 stack - the everyday engine, with Sol as the escalation path when a task outgrows it.

2. Core Features

GPT-5.5-class capability at half price. The core value proposition: previous-flagship quality without previous-flagship costs.

1.5M-token context. The same large window as Sol - long documents, whole codebases, extended conversations.

Effort-level control. Tune reasoning depth per task across supported surfaces, trading tokens for quality.

Tool calling and agent workflows. Function calling across ChatGPT Work, Codex, and the API.

Broad availability. Default for free/Go in Work and Codex; selectable across paid ChatGPT tiers.

Layered safety. The same protections as the family, suited to general workplace use.

3. Technical Specifications

4. Capability Comparison

Against competing mid-tiers: Terra's position is comparable to Gemini's Flash line and Anthropic's Sonnet tier - the "production default" slot in each family. Its differentiators are the 1.5M-token context (larger than most mid-tier rivals) and OpenAI's ecosystem integration depth.

5. Core Advantages

1. The price step-down that enables deployment. Half the cost of the prior flagship converts pilots into production.
2. No context penalty for the middle tier. 1.5M tokens matches the flagship - long-document work isn't reserved for top pricing.
3. Sensible defaults. Already the default in Work/Codex for free and Go users, meaning zero configuration to get started.
4. Escalation path built in. When a task outgrows Terra, Sol is one selector away - same family, same tooling.
5. Effort controls. Cost and latency tunable per task rather than globally.
6. Workplace safety posture. Family-grade safety stack suited to enterprise rollout.

6. Recommended Use Cases

• Daily knowledge work: analysis, drafting, summarization, document processing.
• Coding assistance: completions, explanations, reviews, and test generation where Sol's peak isn't required.
• Office automation: spreadsheet analysis, deck generation, and multi-step workflows in ChatGPT Work.
• Research assistance: literature summarization and structured extraction at 1.5M context.
• Agentic office flows: tool-calling workflows across connected systems.
• Cost-controlled deployment: production workloads where per-token economics are decisive.

7. Example Prompts

1. Document analysis

2. Code review assistance

3. Office workflow

4. Research synthesis

5. Agentic task

8. Selection Recommendations

Choose GPT-5.6 Terra if:

• You want frontier-generation quality at the best price in the family.
• Your work is the everyday middle: analysis, drafting, assistance, automation.
• You need 1.5M-token context without flagship pricing.
• You want zero-config defaults in ChatGPT Work and Codex.

Upgrade to Sol if:

• The task is security research, genomics, complex terminal work, or multi-agent orchestration.
• Accuracy justifies premium token costs.

Downgrade to Luna if:

• You need low-latency, high-concurrency generation at minimum cost.

Sources & Further Reading

• GPT-5.6 - AI Toolset (Chinese overview)
• GPT-5.6 - OpenAI

---

Tier naming, availability, and pricing are as published at review time and subject to change. "Half price" refers to positioning versus the prior generation flagship; verify current rate cards for budgeting.