OpenAI API pricing: 7 models, per token and per task
The list price of every OpenAI model on our list, read from OpenAI’s own pricing page, next to what a passing answer cost on the AIGROW small-business benchmark and what three everyday jobs cost on each.
US dollars per million tokens, at the standard rate. No reseller prices.
- GPT-6 Astra
- $10 / $50
- GPT-5.6 Sol
- $4 / $20
- GPT-5.6 Terra
- $2 / $12
- GPT-5.6 Luna
- $0.20 / $1.20
- GPT-5.5
- $5 / $30
- GPT-5.4 mini
- $0.75 / $4.50
Input / output list price per million tokens.
OpenAI prices the 7 models on our list from GPT-5.6 Luna at $0.20 in and $1.20 out per million tokens to GPT-6 Astra at $10 in and $50 out. On the AIGROW small-business benchmark GPT-5.6 Sol shared the best record, 17 of 20 tasks passed, at the lowest cost: $11.33 per thousand passes.
Per token and per pass
The list price per million tokens, then how many of the benchmark’s 20 office tasks each model passed and what 1,000 passing answers cost, failed runs included.
| Model | Input | Output | Cached input | Context | Tasks passed | 1,000 passes |
|---|---|---|---|---|---|---|
| GPT-6 AstraFlagships | $10.00 | $50.00 | $1.00 | 1.05M | 17 of 20 | $27.91 |
| GPT-5.6 SolFlagships | $4.00 | $20.00 | $0.40 | 1.05M | 17 of 20 | $11.33 |
| GPT-5.6 TerraWorkhorses | $2.00 | $12.00 | $0.20 | 1.05M | 16 of 20 | $6.68 |
| GPT-5.6 LunaSmall and fast | $0.20 | $1.20 | $0.02 | 1.05M | 16 of 20 | $0.677 |
| GPT-5.5Flagships | $5.00 | $30.00 | $0.50 | 1.05M | 17 of 20 | $17.02 |
| GPT-5.4 miniSmall and fast | $0.75 | $4.50 | $0.075 | 400K | 8 of 20 | $3.35 |
| GPT-5.4 nanoSmall and fast | $0.20 | $1.25 | $0.02 | 400K | 6 of 20 | $1.51 |
Where a model ran on a subscription, its cost per pass is worked from the list price above rather than billed usage. gpt-oss-120b has open weights and no list price: OpenAI sells no API for it.
What three everyday jobs cost
1,000 runs of each job at list price. The token counts are ours; the prices are OpenAI’s.
| Model | 1,000 × support reply | 1,000 × document summary | 1,000 × agent task |
|---|---|---|---|
| GPT-6 Astra | $50.00 | $300 | $320 |
| GPT-5.6 Sol | $20.00 | $120 | $128 |
| GPT-5.6 Terra | $10.80 | $62.00 | $70.00 |
| GPT-5.6 Luna | $1.08 | $6.20 | $7.00 |
| GPT-5.5 | $27.00 | $155 | $175 |
| GPT-5.4 mini | $4.05 | $23.25 | $26.25 |
| GPT-5.4 nano | $1.10 | $6.25 | $7.15 |
- Support reply
- A 3,000-token prompt (the ticket, the thread so far and a policy excerpt) and a 400-token answer.
- Document summary
- A 40-page document of about 25,000 tokens in, a 1,000-token summary out.
- Agent task
- Ten steps that read 80,000 tokens between them, 70,000 of them context repeated from step to step and served from cache, and write 3,000.
The small print
What OpenAI’s pricing pages add to the headline rates.
- GPT-6 Astra
- Requests above 272K input tokens bill 2x input and 1.5x output ($20 / $75, cached $2). Cache write $12.50. Batch and Flex 50% off, Fast mode 2x. Data residency adds 10%. Context 1,050,000 with max input 922K.
- GPT-5.6 Sol
- Promotional price since 2026-08-21, 'at least through November 21, 2026'; the previous price was $5 / $30. Requests above 272K input tokens bill 2x input and 1.5x output ($8 / $30). Cache write $5. Batch and Flex 50% off. Data residency adds 10%.
- GPT-5.6 Terra
- OpenAI calls it the mini tier of the 5.6 family; tiered mid here on price. Price cut 20% on 2026-07-30. Requests above 272K input tokens bill 2x input and 1.5x output ($4 / $18). Cache write $2.50. Batch and Flex 50% off. Data residency adds 10%.
- GPT-5.6 Luna
- Nano tier of the 5.6 family. Price cut 80% on 2026-07-30. Requests above 272K input tokens bill 2x input and 1.5x output ($0.40 / $1.80). Cache write $0.25. Batch and Flex 50% off. Data residency adds 10%.
- GPT-5.5
- Previous flagship. Requests above 272K input tokens bill 2x input and 1.5x output ($10 / $45). Cached price applies only with extended prompt caching. Batch and Flex 50% off. Data residency adds 10%.
- GPT-5.4 mini
- Batch 50% off. Context 400,000 with max input 272K.
- GPT-5.4 nano
- Batch 50% off.
Head to head
The comparisons with a OpenAI model on one side, each with both prices and the three jobs costed on each.
- GPT-6 Astra vs Claude Fable 5.1
- GPT-6 Astra vs Gemini 3.1 Pro (preview)
- GPT-6 Astra vs GPT-5.6 Sol
- GPT-5.5 vs GPT-5.6 Sol
- Claude Opus 5 vs GPT-5.6 Sol
- GPT-5.6 Sol vs Gemini 3.1 Pro (preview)
- Grok 4.7 vs GPT-5.6 Sol
- DeepSeek V4 Pro vs GPT-5.6 Sol
- Muse Spark 1.3 vs GPT-5.6 Sol
- Claude Sonnet 5 vs GPT-5.6 Terra
- GPT-5.6 Terra vs Gemini 3.8 Flash
- GPT-5.6 Sol vs GPT-5.6 Terra
- Kimi K2.6 vs GPT-5.6 Terra
- Llama 4 Maverick vs gpt-oss-120b
- gpt-oss-120b vs DeepSeek V4.1 Flash
- GPT-5.6 Luna vs Gemini 3.5 Flash-Lite
- GPT-5.6 Luna vs Gemini 3.1 Flash-Lite
- Claude Haiku 4.5 vs GPT-5.6 Luna
- Claude Haiku 4.5 vs GPT-5.4 mini
- GPT-5.4 mini vs GPT-5.6 Luna
- GPT-5.4 nano vs GPT-5.6 Luna
- DeepSeek V4.1 Flash vs GPT-5.6 Luna
- Mistral Small 4 vs GPT-5.6 Luna
Where the prices come from
Read on 22 September 2026. Each price is the standard rate for the shortest context band, with no batch or priority discount; long-context surcharges, batch rates and promotions are in each model’s small print. No price was taken from a reseller.
Pay a quarter of list price
AIGROW API credit is metered at the makers’ list prices, like the ones above, and $25 buys $100 of usage. The credit page lists the models it covers.
Prices checked 22 September 2026