benchrPricing breakdown
| Tier | Rate / 1M tokens |
|---|---|
| Standard input | $0.25 |
| Standard output | $2.00 |
| Cached input | $0.025 |
| Batch input | $0.125 |
| Batch output | $1.00 |
| Context window | 400,000 tokens |
| Maximum output | 128,000 tokens |
Context window and maximum output are separate
OpenAI lists a 400,000-token context window and a separate 128,000-token maximum output. The first is the total request window; the second caps generated output. Output still consumes part of the overall context budget, so 400,000 tokens of input plus 128,000 tokens of output is not the correct interpretation. Reserve room for the requested output and any other message or tool content.
What OpenAI publishes — and what it does not
The checked GPT-5 Mini model page publishes price, context, maximum output, supported endpoints, and features. It does not list a SWE-bench result or a numeric tokens-per-second claim. Benchr leaves those formal fields blank instead of presenting an unsourced estimate.
For coding, run representative repository issues and review the patches. For classification, extraction, routing, and customer-service workloads, evaluate accuracy, escalation rate, latency, and cost on a labelled sample from your own traffic.
Where the published price may fit
At $0.25/1M input, GPT-5 Mini can be economical for pipelines making many small, well-specified calls. Actual throughput varies with region, load, output length, and request settings, and the checked provider page does not state a numeric rate. Measure latency on your deployment before sizing capacity.
Batch API: the $0.125 route
GPT-5 Mini via the OpenAI Batch API lists at $0.125/1M input and $1/1M output. For offline document processing, data labeling, eval generation, or other pipelines that can tolerate the provider's current batch turnaround, compare that published rate with the alternatives available when you deploy. At $0.125 per million, a 1B-token monthly input workload costs $125 before output charges.
Cost scenarios
At the listed rates, 100M input + 20M output tokens per month costs $25 + $40 = $65/month. Using the comparison rates recorded for this snapshot, GPT-5 totals $325 and Claude Haiku 4.5 totals $200 for the same token mix. This is rate-card arithmetic only: current prices, caching, tool charges, retries, and task accuracy can change total workload cost.
Use-case fit
Test for: Intent classification; sentiment analysis; structured extraction; customer-service routing; content moderation; and batch document labelling where the listed price materially affects unit economics.
Do not fill the missing benchmark with a guess: the provider's absent SWE-bench result does not prove suitability or unsuitability for a task. Use a representative quality evaluation. Treat the published 400,000-token context and 128,000-token maximum output as separate constraints.
Decision checklist
Before choosing GPT-5 Mini, run a labelled quality evaluation on your actual task distribution, including coding or complex cases if your traffic contains them. Compare accuracy, escalation rate, latency, and total cost with the alternatives; no provider-published benchmark on this page defines the next quality tier.
For batch workloads: calculate monthly token volume and compare $0.125 batch vs $0.25 standard. If turnaround time allows async execution, the batch route cuts costs in half.