OpenAI
GPT-5.4 pricing
OpenAI's next-generation flagship. 1M context and improved reasoning over GPT-5, priced 2x higher on input.
Input
$2.50/ 1M tok
Output
$15.00/ 1M tok
- Context window
- 1.1M
- Max output
- 128K
- Cached input
- $0.250 / 1M
- Verified
- 2026-04-06
GPT-5.4 is OpenAI's current-generation flagship - $2.50 per 1M input and $15 per 1M output, with a 1M context window that doubles the GPT-5 ceiling. Both sides of the bill ran up versus GPT-5 ($1.25 / $10), so moving traffic from 5 to 5.4 is a real cost increase; justify it on capability, not defaults.
The cached input rate at $0.25 per 1M gives a 90% discount, which matters a lot for workflows with long static prefixes (system prompts, reference documents). OpenAI applies prompt caching automatically once a prefix repeats within a 5-minute window, so you get the savings without wiring anything up.
Calcis counts GPT-5.4 input tokens with o200k_base (tiktoken), which is the exact tokenizer OpenAI uses for billing - so the token count you see matches what lands on your invoice.
Estimate your cost on GPT-5.4
Paste your prompt into the estimator, pick GPT-5.4, and see the exact dollar cost - input tokens counted with the provider's own tokenizer, output tokens predicted by our regression model.
Frequently asked
- How much does GPT-5.4 cost per request?
- A 1,000-token prompt with a 500-token reply costs about $0.01 ($0.0025 input + $0.0075 output). The 8:1 output-to-input ratio means most of your bill comes from the response side.
- Is GPT-5.4 worth 2x the price of GPT-5?
- Depends on the workload. For complex reasoning, long-context document work, or tasks where you measured a quality gap on GPT-5, yes. For chat, summary, and routing, GPT-5 at $1.25 / $10 is usually indistinguishable.
- Does GPT-5.4 have a long-context surcharge?
- No. Unlike Gemini 2.5 Pro, OpenAI bills flat per-token rates across the full 1M context window. No threshold to worry about.
- What's the cached input discount on GPT-5.4?
- $0.25 per 1M cached tokens - a 90% discount on the standard $2.50 input rate. Applied automatically when prompt prefixes repeat within 5 minutes.
Pricing verified 2026-04-06 from the provider's rate card.