GPT-5 mini
GPT-5 mini — OpenAI's most cost-efficient model for high-volume production workloads.
| Type | Price / 1M tokens | Notes |
|---|---|---|
| Input | $0.25 | Per 1M input tokens |
| Output | $2.00 | Per 1M output tokens |
| Cache read | $0.025 | Cached prefix read-back |
| Batch API | $0.13 in / $1.00 out | ~50% off, async (up to 24h) |
At $0.25/M input and $1/M output, GPT-5 mini is the cheapest capable model in OpenAI's lineup and often the right choice for production at scale.
Best for
- High-volume classification and extraction
- Simple chatbots and FAQ
- Routing layer decisions
- Batch data processing
Not ideal for
- Complex multi-step reasoning
- Nuanced creative writing
- Tasks requiring strict instruction adherence
Real cost scenarios
| Scenario | Est. monthly cost | Breakdown |
|---|---|---|
| 10M requests/month — 500 in, 200 out | $3,250 | $1,250 input + $2,000 output |
| Email classification — 5M emails/month | $750 | 300 tokens in, 50 out |
Estimates with typical token distributions. Use the Prompt Cost Calculator for your exact workload.
How to cut costs on GPT-5 mini
- GPT-5 mini is already the cheapest capable OpenAI model. Use batch API for additional 50% savings on async workloads.
- OpenAI's automatic prompt caching applies to GPT-5 mini — no configuration needed, 50% off cached tokens.
Calculate your GPT-5 mini costs
Other OpenAI models
Compare GPT-5 mini
Frequently Asked Questions
Is GPT-5 mini the same quality as GPT-5.4?
No. GPT-5 mini scores 8-12% lower on complex reasoning benchmarks. For simple tasks (classification, extraction, FAQ) the quality difference is imperceptible. For complex tasks (multi-step reasoning, code review, nuanced analysis), you'll notice quality gaps. Evaluate on your specific task before choosing.
When should I use GPT-5 mini vs Claude Haiku 4.5?
They are priced comparably and similar in capability. GPT-5 mini has slightly better OpenAI-specific structured output adherence. Claude Haiku has 200K context vs 128K. Choose based on which ecosystem you're already in, or run an A/B eval on your specific task.
Pricing verified 2026-08-19 from OpenAI official pricing. All prices in USD per 1M tokens. See methodology and price changelog.