GPT-4 Turbo pricing
OpenAI · list prices as of 2026-09-29
| Input | $10 per 1M tokens |
|---|---|
| Output | $30 per 1M tokens |
| Batch / async | 50% off |
| Context window | 128K tokens |
| Longest answer | 4K tokens |
Uses the older cl100k_base encoding, so the same text costs more tokens here.
What a call costs
| Typical call | Input tokens | Output tokens | Per call | Per month at 1,000 a day |
|---|---|---|---|---|
| Chat reply | 1,000 | 300 | $0.0190 | $570.00 |
| Document summary | 6,000 | 400 | $0.0720 | $2,160 |
| Code change | 3,000 | 1,200 | $0.0660 | $1,980 |
No caching or batch discount. Your own prompt will differ: count it exactly below.
Price your own prompt on GPT-4 TurboSimilarly priced models
- Claude Fable 5 (Anthropic): 1.3× more
- Claude Fable 5.1 (Anthropic): 1.3× more
- Claude Mythos 5 (Anthropic): 1.3× more
- Claude Mythos 5.1 (Anthropic): 1.3× more
- GPT-5.5 (OpenAI): 1.4× cheaper
Questions
How much does GPT-4 Turbo cost?
GPT-4 Turbo costs $10 per million input tokens and $30 per million output tokens, as of 2026-09-29.
How much is one GPT-4 Turbo call?
A call with 1,000 input tokens and a 300-token answer costs about $0.0190, or $570.00 a month at 1,000 calls a day.
How big is the GPT-4 Turbo context window?
128,000 tokens.