GPT-5.6 Terra: Everyday Work, Lower Cost, and When It's Enough Instead of Sol
GPT-5.6 Terra is OpenAI's mid-tier model in the GPT-5.6 family, built to handle everyday work at a fraction of the cost of the flagship Sol model.
........
Terra sits between Sol (flagship) and Luna (fastest, cheapest) in OpenAI's three-tier GPT-5.6 lineup.
At launch, OpenAI described Terra as matching GPT-5.5 while costing half as much.
On July 30, 2026, OpenAI cut Terra's price by 20%, on top of an 80% cut for Luna.
Terra supports a context window of roughly 1 million tokens and up to 128,000 tokens of output.
Terra isn't selectable in standard ChatGPT chat; it's available in ChatGPT Work, Codex, and via the API.
··········
WHAT GPT-5.6 TERRA IS.
Terra is the "balanced" tier of the GPT-5.6 family, positioned for workloads where cost matters as much as capability.
OpenAI describes Terra as designed for workloads that balance intelligence and cost, roughly corresponding to the mini tier from earlier GPT-5 generations.
It sits below Sol, the flagship reasoning model, and above Luna, the fastest and cheapest tier.
At launch, OpenAI said Terra offered performance competitive with GPT-5.5 while costing half as much to run.
··········
PRICING: THEN AND NOW.
Terra has already gone through one price cut since its general availability launch.
At general availability, Terra was priced at $2.50 per million input tokens and $15 per million output tokens.
On July 30, 2026, OpenAI reduced Terra's price by 20%, bringing it to $2 per million input tokens and $12 per million output tokens.
Cache reads are billed separately at $0.20 per million tokens, and cache writes at $2.50 per million tokens.
Web search calls are billed at $10 per 1,000 calls.
··········
Terra pricing: launch vs. current
Metric | At launch | After July 30, 2026 cut |
|---|---|---|
Input tokens | $2.50 / million | $2.00 / million |
Output tokens | $15 / million | $12.00 / million |
Price change | — | -20% |
··········
SPECS AND REASONING CONTROLS.
Terra supports a large context window and adjustable reasoning depth.
The context window runs to roughly 1,050,000 tokens, with up to 128,000 tokens of output per response.
The reasoning.effort parameter supports six levels: none, low, medium (the default), high, xhigh, and max.
Terra accepts tools and tool_choice for function calling, and supports structured outputs through a JSON schema in response_format.
It also accepts files such as PDFs, images, and text as input.
··········
WHERE TERRA IS AVAILABLE.
Terra is not one of the models people pick directly inside a standard ChatGPT conversation.
In ChatGPT, Sol powers the Instant and reasoning options on paid plans, and Sol Pro powers the Pro tier.
Terra and Luna are not selectable in those standard chats.
Depending on the plan, Terra and Luna are available in ChatGPT Work and in Codex, OpenAI's coding-agent product.
Developers can also access Terra directly through the OpenAI API.
··········
REAL-WORLD RESULTS.
OpenAI has published early customer feedback on how Terra performs in production.
Notion reported that Terra was a strong fit for everyday work in its personal agent, including workspace Q&A and scoped tasks where latency matters, delivering comparable quality to GPT-5.5 at half the cost per task and in 60% less time.
Another customer noted that Terra and Luna led on cost-efficiency across their internal coding benchmarks, ahead of the flagship Sol tier.
··········
WHEN TERRA IS ENOUGH — AND WHEN TO REACH FOR SOL.
Terra is built for volume: everyday requests where speed and cost matter more than squeezing out the last bit of reasoning depth.
For customer support, workspace Q&A, routine coding tasks, and scoped agent work, Terra was designed to match GPT-5.5-level quality at roughly half the cost.
For high-stakes reasoning, complex multi-step problems, or research-grade work, Sol remains OpenAI's recommended tier.
A common pattern among early adopters is running most traffic on Terra and escalating only the harder edge cases to Sol.
··········
AVAILABILITY AND ACCESS.
GPT-5.6 Terra is generally available today through the OpenAI API, ChatGPT Work, and Codex.
It is not available as a selectable option in standard ChatGPT Free, Plus, or Pro conversations, where Sol and Sol Pro handle everyday and advanced requests respectively.
Multiple hosting platforms, including OpenRouter, also offer access to Terra with routing options for price, speed, or tool-calling accuracy.
··········
·····
FOLLOW US FOR MORE.
·····
·····
DATA STUDIOS
·····
