PRICING
Plan with the unit in view.
Compare indicative prices and their public references before choosing a model or service. No entry on this page is an active rate, executable quote, invoice, or availability commitment.
Use the local catalog estimator for text-token planning. It sends no model request and does not create an order.
MODEL & SERVICE PRICING
Clear units.
Transparent pricing.
Compare indicative model and service prices for your workload. These are planning estimates, not billable rates or a quote; listed services are not available for invocation.
Indicative price book pacinfrax-draft-2026-10-03-v1 · USD · Public references checked 2026-10-03 · Selected model and service estimates. Final prices depend on licensing, deployment requirements and service terms.
Explore the catalog and estimate a text workload →
Text inference
| Model reference | Unit | Indicative input | Indicative output | Public reference input / output |
|---|---|---|---|---|
| GPT-OSS 120B | USD / 1M input or output tokens | $0.13875 | $0.555 | $0.15 / $0.60 |
| Llama 3.3 70B Instruct Turbo | USD / 1M input or output tokens | $0.962 | $0.962 | $1.04 / $1.04 |
| Qwen3.5 9B | USD / 1M input or output tokens | $0.15725 | $0.23125 | $0.17 / $0.25 |
Developer services
| Service reference | Unit | Indicative price | Public reference |
|---|---|---|---|
| Shared Filesystem | USD / GiB-month | $0.148 | $0.16 |
| Code Sandbox — CPU | USD / vCPU-hour | $0.041255 | $0.0446 |
| Code Sandbox — RAM | USD / GiB RAM-hour | $0.0137825 | $0.0149 |
| Code Interpreter | USD / session (60 minutes) | $0.02775 | $0.03 |
Not billable. Reference model IDs are not PacinfraX callable IDs. Region, revision, runtime, throughput and commercial rights remain unverified. Sandbox CPU and memory are separate components; totals depend on both. Storage, network, support, taxes and other charges need explicit terms. These indicative values are not connected to gateway accounting.