OpenAI Cuts GPT-5.6 Terra and Luna Prices and Adds Sol Fast Mode
Terra falls to $2/$12 per million tokens, Luna to $0.20/$1.20, while Sol gains a premium low-latency API path.
OpenAI changed the economics and service tiers of GPT-5.6 on July 30, cutting the API prices of its Terra and Luna models while introducing a faster processing option for Sol. The official product update says the lower model prices also reduce how quickly paid ChatGPT Work and Codex users consume usage credits.
What changed
GPT-5.6 Terra now costs $2 per million input tokens and $12 per million output tokens in the API. GPT-5.6 Luna now costs $0.20 per million input tokens and $1.20 per million output tokens. OpenAI describes those changes as reductions of 20% for Terra and 80% for Luna.
The company says ChatGPT and Codex subscription prices and quota budgets have not changed. Instead, Terra and Luna usage now consumes fewer credits inside paid subscriptions. Availability is otherwise unchanged: both models remain offered through ChatGPT Work, Codex and the OpenAI API, with Terra available to Free and Go users and model choice available on higher plans.
OpenAI also introduced Fast mode for GPT-5.6 Sol in the API. It replaces Priority Processing, charges twice the Standard processing price and is advertised as delivering up to 2.5 times faster performance without changing model intelligence. Existing requests tagged for priority processing remain compatible and are routed through Fast mode automatically.
Why it matters
The update changes deployment decisions more than model capability. Teams running high-volume classification, document processing or routine agent steps can now buy the same Terra and Luna models at lower unit prices. Developers with latency-sensitive Sol workloads gain a clearly priced premium path, while existing priority integrations do not require an immediate request-format change.
The percentage cuts should not be read as proof that every workload becomes cheaper by the same amount. Real cost still depends on input length, generated output, caching, retries and how often an agent invokes tools. OpenAI’s page describes its own efficiency gains and customer experience; it does not provide an independent comparison across representative production workloads.
Status
Confirmed company announcement. Internal confidence is medium because pricing and availability are documented by OpenAI, while the speed and efficiency characterizations remain vendor claims without independent testing in the fetched evidence.
Sources
Update note: Last reviewed 2026-07-31. We will revise this post if OpenAI changes the prices, plan-credit treatment or Fast mode terms.
Sources
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.