GPT-5.6 Terra

gpt-5.6-terraby OpenAI · gpt-terra series

Balanced GPT-5.6 model for capable, cost-efficient everyday work

Modalities
Context window
1.05M tokens
Max output
128K tokens
Release date
2026-07-09
Knowledge cutoff
2026-02-16
Capabilities
函数工具推理强度档位JSON Schema 输出

Declarations match gateway routing: unsupported parameters are rejected before the request runs, at no cost

Pricing

USD / 1M tokens · the listed price is the settled price; streaming and non-streaming cost the same, failed requests are auto-refunded

Billing itemUnioAPI priceList price
Standard
Uncached input$0.14$2.00
Cache read$0.014$0.20
Cache creation30 min$0.175$2.50
Output$0.84$12.00
Priority · OpenAI expedited lane
Uncached input$0.28$4.00
Cache read$0.028$0.40
Cache creation30 min$0.35$5.00
Output$1.68$24.00
Long-context tiered pricing

When a single request's total input exceeds 272K tokens, the whole request is billed at input ×2 and output ×1.5 (i.e. input $0.28, output $1.26 per 1M).

Quick start

OpenAI-compatible protocol — point your SDK's base URL at UnioAPI, no code changes

curl https://api.unioapi.com/v1/chat/completions \
  -H "Authorization: Bearer $UNIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-5.6-terra","messages":[{"role":"user","content":"Hello"}]}'

FAQ

Do these prices match my actual bill?
Yes. This page reads the same data source as the billing system, so the listed price is the settled price. Price changes apply immediately and only to subsequent requests.
What is the Priority tier?
Priority is OpenAI's expedited lane and is only available on OpenAI models. Shorter queues and faster first tokens, billed at its own rates. Select it with service_tier: "priority".
How is this different from the official API?
Fully protocol-compatible with identical capabilities. The difference: lower prices, and failed requests are automatically refunded.

Start calling GPT-5.6 Terra

Create an API key after signing up — pay per token, no minimum spend, failed requests refunded.

Get started