GPT-6 API pricing
OpenAI's current generation, released in September 2026 in three sizes: Astra ($10 / $50), Sol ($2 / $10) and Luna ($0.10 / $0.50) per 1M tokens, all with a 1.05M-token context window.
GPT-6 Astra
New 3 SepWhat it costs in practice
GPT-6 Astra at list prices, no batch discount.
| 1,000 chat replies1,500 tokens in and 300 out each | $30.00 |
|---|---|
| Summarising 100 long reports20,000 tokens in and 800 out each | $24.00 |
| A month of a coding agent400 tasks, 85% of input cached | $2,163 |
The coding-agent example assumes 85% of input is served from cache, typical for agents that resend the same context each step. Model your own workload
Details
gpt-6-astra- Broadly available from 4 September 2026.
Checked 27 Sep 2026 against OpenAI pricing and OpenAI model page .
Other GPT-6 models
Prompts longer than 272K tokens are billed at 2× input and 1.5× output.
| Model | Input / 1M | Output / 1M | Cached input | Context | Released |
|---|---|---|---|---|---|
| $2.00 | $10.00 | $0.20 | 1.05M | 22 Sep 2026 | |
| $0.10 | $0.50 | $0.010 | 1.05M | 22 Sep 2026 |
Version details
GPT-6 Sol
New 22 Sep$2.00 / $10.00 per 1M tokenscached $0.20 · batch 50% off · 1.05M context
Checked 27 Sep 2026 against OpenAI pricing and OpenAI model page .
GPT-6 Luna
New 22 Sep$0.10 / $0.50 per 1M tokenscached $0.010 · batch 50% off · 1.05M context
Checked 27 Sep 2026 against OpenAI pricing .
Compare GPT-6 Astra
Price history
All changesOpenAI pricing rules
- Batch and Flex processing cost 50% less.
- For GPT-5.4 and later, prompts longer than 272K tokens are billed at 2× the input price and 1.5× the output price.
- Cached input is billed at about a tenth of the input price.
Questions
How much does GPT-6 Astra cost?
As of 27 Sep 2026, GPT-6 Astra costs $10 per million input tokens and $50 per million output tokens, and $1 per million cached input tokens. Batch requests cost 50% less. Prices checked 27 Sep 2026.
What is the cheapest GPT-6 model?
GPT-6 Luna, at $0.10 / $0.50 per million input / output tokens.
What is GPT-6 Astra's context window?
1.05M tokens.
Does GPT-6 Astra support prompt caching?
Yes. Repeated input read from the cache costs $1 per million tokens, 90% less than normal input.