A8gent
GlossaryAgent concepts

What is GPT-5.6 Terra?

Also called: gpt-5.6-terra, gpt 5.6 terra, gpt-5.6-terra pricing, openai terra model

GPT-5.6 Terra is OpenAI's balanced tier in the GPT-5.6 family, priced at $2.00 per million input tokens and $12.00 per million output tokens with the same 1,050,000-token context window as the frontier model.

Written by A8gent
Contents
  1. At a glance
  2. What GPT-5.6 Terra is
  3. What it costs
  4. Where Terra is the right default
  5. What you should know
  6. FAQ
  7. Related terms

At a glance

Model ID
gpt-5.6-terra
Input price
$2.00 per 1M tokens (checked August 2026)
Cached input price
$0.20 per 1M tokens
Output price
$12.00 per 1M tokens
Context window
1,050,000 tokens
Max output
128,000 tokens
Positioned as
The GPT-5.6 model that balances intelligence and cost
Versus Sol
40% of the input price and 40% of the output price, same context window

What GPT-5.6 Terra is

gpt-5.6-terra is the middle model ID in OpenAI's GPT-5.6 family. OpenAI describes it as the GPT-5.6 model that balances intelligence and cost. It sits between gpt-5.6-sol at the frontier and gpt-5.6-luna at the cheap end.

As published in August 2026 it carries the same 1,050,000-token context window and 128,000 maximum output tokens as its siblings. That is worth stating plainly because it is unusual: on most vendor lineups the cheaper tiers get a smaller window, and here they do not. Choosing Terra over Sol costs you capability, not room.

What it costs

OpenAI's pricing page in August 2026 lists gpt-5.6-terra at $2.00 per million input tokens, $0.20 per million cached input tokens and $12.00 per million output tokens.

Against Sol that is 40% of the input price and 40% of the output price - a consistent 2.5x saving on both meters. The cached input rate holds the same 10:1 ratio to standard input, so prefix caching is worth exactly as much here proportionally as it is on the frontier tier.

Where Terra is the right default

Terra is the tier most production agents should start on. The pattern that works is: build on Terra, assemble an eval set from real failures, and then route only the cases that fail to Sol. That gives you frontier quality on the 5% of traffic that needs it and mid-tier cost on the other 95%, which is a far better outcome than picking one model for everything.

The arithmetic on a 10,000-conversation month at six turns, 5,700 input and 400 output tokens per turn: Terra costs about $1,400 uncached against Sol's $3,400 for identical traffic. With prefix caching both fall, and the ratio holds.

What you should know

Do not choose a tier from a benchmark table. The only measurement that predicts your outcome is your own task, run against your own inputs, scored against your own criteria - which is what an eval set is for. Vendors' relative rankings are averages over benchmarks that may share nothing with what you are doing.

Prices and model IDs change. This entry was checked in August 2026 against OpenAI's pricing page; verify there before committing a budget. And remember the platform meter is separate - see execution vs task vs operation.

GPT-5.6 Terra FAQ

How much does gpt-5.6-terra cost?

OpenAI listed $2.00 per million input tokens, $0.20 per million cached input tokens and $12.00 per million output tokens in August 2026. That is 40% of the frontier model's price on both the input and output meters.

Does GPT-5.6 Terra have a smaller context window than Sol?

No. Terra carries the same 1,050,000-token context window and 128,000-token output ceiling as Sol and Luna. On this lineup the tiers differ in capability and price, not in how much you can put in front of them.

Should I default to Terra or Sol?

Start on Terra and escalate. Build an eval set from real failures, route only the failing cases to Sol, and you get frontier quality where it matters at mid-tier cost everywhere else. Picking one model for all traffic is almost always the more expensive answer.

How much cheaper is Terra than Sol in practice?

A 2.5x saving on both meters. On 10,000 conversations of six turns at 5,700 input and 400 output tokens per turn, Terra is roughly $1,400 uncached against Sol's $3,400 for identical traffic. Prefix caching reduces both proportionally.

Is Terra good enough for agent tool calling?

That is exactly the question an eval set answers, and it depends on how many tools you expose and how ambiguous the choice between them is. Test it with your real tool schema rather than a toy one, because tool-selection accuracy degrades with the number and similarity of tools far more than with the prompt.

Primary source

OpenAI - API pricing

Checked August 2026. Anything priced or rate-limited moves, so verify against the source before you budget.

Want this built rather than explained?

Knowing what gpt-5.6 terra means is one thing; having it running in your business is another. We scope one workflow at a time, quote a fixed price in writing after a free consult, build it, and hand it over with documentation you own.