GPT-5.6 Terra

Price history

Prompts above 272K input tokens have higher rates. Cache writes and tool calls can add charges beyond these standard token prices.

When live checks began

These are the earliest and latest successful checks we can prove from our retained polling records. Reconstructing an older price never moves these timestamps backward.

A baseline of these monitoring records was preserved on 2026-09-18. The current checks below may be newer.

Launches & price changes

A reconstruction from dated sources. The effective date, publication date, and date we added the evidence are recorded separately. Coverage is incomplete; an absent event does not prove nothing changed.

1. 2026-07-30Price changestandard

   July 30 permanent cut: Terra 20%, Luna 80%, applied to their published launch rates; cache reads retain the 90% discount.

   USD / 1M tokens: input $2 · output $12 · cached input $0.2

   Published 2026-07-30 · Added to our history 2026-09-18

   • OpenAI — July 30 permanent cut: Terra 20%, Luna 80%, applied to their published launch rates; cache reads retain the 90% discount.
   • OpenAI — July 30 permanent cut: Terra 20%, Luna 80%, applied to their published launch rates; cache reads retain the 90% discount.
2. 2026-07-09API launchstandard

   GPT-5.6 API general availability. Standard rates exclude long-context premiums; cache_price is cache read. Cache writes cost 1.25 times uncached input.

Benchmarks

Results depend on reasoning effort, tools, and evaluation setup. Check the original report before comparing scores across models.