OpenAI API prices: the GPT-5.6 line has long-context rates too
2026/09/02
Per million tokens, GPT-6 is priced at $10 input and $50 output on gpt-6-astra, $2 and $10 on gpt-6-sol, and $0.10 and $0.50 on gpt-6-luna[1].
All three have a second set of rates for long prompts, with input and cache doubling and output rising by half[1]. That makes Astra $20 and $75, Sol $4 and $15, and Luna $0.20 and $0.75[1]. Sol and Luna switch to these rates once a prompt passes 272K input tokens[2,3].
GPT-6 Sol costs exactly half of gpt-5.6-sol on both input and output[1]. GPT-6 Luna is half of gpt-5.6-luna on input and less than half on output, at $0.50 against $1.20[1].
The GPT-5.6 line has three main models at different prices[1]. Per million tokens, gpt-5.6-sol is $4 input and $20 output, gpt-5.6-terra is $2 and $12, and gpt-5.6-luna is $0.20 and $1.20[1].
These three carry long-context rates as well: sol goes to $8 input and $30 output, terra to $4 and $18, and luna to $0.40 and $1.80[1]. Input doubles and output rises by half, the same shape as GPT-6[1].
Cached input is one tenth of the input rate on every GPT-6 and GPT-5.6 model, and a cache write is 1.25 times it[1]. Batch and Flex are half the standard rates, and Fast mode is twice them[1].
A fourth GPT-5.6 model, gpt-5.6-cyber, sits outside the pattern at $12.50 input and $75 output, with cached input at $1.25 and cache writes at $15.625[1]. Only standard rates are listed for it, with no long-context or batch rates[1].
The price on gpt-5.6-sol is a promotional one, held at least through 21 November 2026[1]. The older gpt-5.5 remains more expensive than sol at $5 and $30[1].
