Servicio de modelos fundacionales propietarios
Utilice modelos fundacionales propietarios de última generación para las necesidades de cargas de trabajo de inferencia en tiempo real y por lotes. Esto le permite crear aplicaciones de forma rápida y sencilla que aprovechan modelos de IA generativa propietarios de alta calidad de varios proveedores directamente en la plataforma Databricks, sin la necesidad de interactuar adicional y separadamente con otros proveedores.
* Para los clientes de Azure, si tienes un compromiso de Azure con Databricks, Databricks puede poner a disposición este servicio como un servicio ADI que se integra con Azure Databricks. El servicio ADI es vendido y facturado por Databricks. Contactar a Ventas para obtener acceso.
1. Azure Databricks, como un servicio de origen en Microsoft Azure, ofrece facturación y soporte unificados por parte de Microsoft.
1. El nivel Premium en Azure Databricks corresponde al nivel Enterprise en AWS y GCP
Proprietary Foundation Model Serving DBU rates
| Model | Endpoint type | Context Length | Pay Per Token | Batch Inference | |||
|---|---|---|---|---|---|---|---|
| Input | Output | Cache writes | Cache reads | ||||
| DBU / 1M Tokens | DBU / 1M Tokens | DBU / 1M Tokens | DBU / 1M Tokens | DBU / hour | |||
| OpenAI | |||||||
| GPT 5.5 | Global | Short | 71.429 | 428.571 | 71.429 | 7.143 | 214.286 |
| In-geo | 78.572 | 471.428 | 78.572 | 7.857 | 235.715 | ||
| Global | Long | 142.857 | 642.857 | 142.857 | 14.286 | 214.286 | |
| In-geo | 157.143 | 707.143 | 157.143 | 15.714 | 235.715 | ||
| GPT 5.4 Pro | Global | Short | 428.571 | 2,571.429 | 428.571 | 42.857 | 1,142.857 |
| In-geo | 471.428 | 2,828.572 | 471.428 | 47.143 | 1,257.143 | ||
| Global | Long | 857.142 | 3,857.144 | 857.142 | 85.714 | 1,142.857 | |
| In-geo | 942.856 | 4,242.858 | 942.856 | 94.286 | 1,257.143 | ||
| GPT 5.4 | Global | Short | 35.714 | 214.286 | 35.714 | 3.571 | 192.857 |
| In-geo | 39.285 | 235.715 | 39.285 | 3.929 | 212.143 | ||
| GPT 5.4 | Global | Long | 71.428 | 321.429 | 71.428 | 7.143 | 192.857 |
| In-geo | 78.571 | 353.572 | 78.571 | 7.857 | 212.143 | ||
| GPT 5.4 mini | Global | All Lengths | 10.714 | 64.286 | 10.714 | 1.071 | 107.143 |
| In-geo | 11.786 | 70.714 | 11.786 | 1.179 | 117.857 | ||
| GPT 5.4 nano | Global | All Lengths | 2.857 | 17.857 | 2.857 | 0.286 | 71.429 |
| In-geo | 3.143 | 19.643 | 3.143 | 0.314 | 78.571 | ||
| GPT 5.2/5.3 Codex | Global | All Lengths | 25.000 | 200.000 | 25.000 | 2.500 | n/a |
| In-geo | 27.500 | 220.000 | 27.500 | 2.750 | n/a | ||
| GPT 5.2 | Global | All Lengths | 25.000 | 200.000 | 25.000 | 2.500 | 184.286 |
| In-geo | 27.500 | 220.000 | 27.500 | 2.750 | 202.714 | ||
| GPT 5.1 | Global | All Lengths | 17.857 | 142.857 | 17.857 | 1.786 | 131.429 |
| In-geo | 19.643 | 157.143 | 19.643 | 1.965 | 144.571 | ||
| GPT 5.1 Codex Max | Global | All Lengths | 17.857 | 142.857 | 17.857 | 1.786 | n/a |
| In-geo | 19.643 | 157.143 | 19.643 | 1.965 | n/a | ||
| GPT 5 | Global | All Lengths | 17.857 | 142.857 | 17.857 | 1.786 | 131.429 |
| In-geo | 19.643 | 157.143 | 19.643 | 1.965 | 144.571 | ||
| GPT 5 mini | Global | All Lengths | 3.571 | 28.571 | 3.571 | 0.357 | 71.429 |
| In-geo | 3.929 | 31.429 | 3.929 | 0.393 | 78.571 | ||
| GPT 5.1 Codex Mini | Global | All Lengths | 3.571 | 28.571 | 3.571 | 0.357 | n/a |
| In-geo | 3.929 | 31.429 | 3.929 | 0.393 | n/a | ||
| GPT 5 nano | Global | All Lengths | 0.714 | 5.714 | 0.714 | 0.071 | 53.571 |
| In-geo | 0.786 | 6.286 | 0.786 | 0.078 | 58.929 | ||
Proprietary Foundation Model Serving DBU rates
| Model | Endpoint type | Context Length | Pay Per Token | Batch Inference | |||
|---|---|---|---|---|---|---|---|
| Input | Output | Cache writes | Cache reads | ||||
| DBU / 1M Tokens | DBU / 1M Tokens | DBU / 1M Tokens | DBU / 1M Tokens | DBU / hour | |||
| Anthropic | |||||||
| Claude Opus 4.5 / 4.6 / 4.7 / 4.8 | Global | All Lengths | 71.429 | 357.143 | 89.286 | 7.143 | 178.571 |
| In-geo | 78.571 | 392.857 | 98.214 | 7.857 | 196.429 | ||
| Claude Opus 4 / 4.1 | Global/In-geo | All Lengths | 214.286 | 1,071.429 | 267.857 | 21.429 | 514.286 |
| Claude Sonnet 4.5 / 4.6 | Global | All Lengths | 42.857 | 214.286 | 53.571 | 4.286 | 214.286 |
| In-geo | 47.143 | 235.715 | 58.928 | 4.715 | 235.715 | ||
| Claude Sonnet 4 | Global/In-geo | All Lengths | 42.857 | 214.286 | 53.571 | 4.286 | 214.286 |
| Claude Haiku 4.5 | Global | All Lengths | 14.286 | 71.429 | 17.857 | 1.429 | 114.286 |
| In-geo | 15.715 | 78.572 | 19.643 | 1.572 | 125.714 | ||
Proprietary Foundation Model Serving DBU rates
| Model | Endpoint type | Context Length | Pay Per Token | Batch Inference | |||
|---|---|---|---|---|---|---|---|
| Input | Output | Cache writes | Cache reads | ||||
| DBU / 1M Tokens | DBU / 1M Tokens | DBU / 1M Tokens | DBU / 1M Tokens | DBU / hour | |||
| Gemini 3.5 Flash | Global | All Lengths | 26.786 | 160.714 | 26.786 | 2.679 | 196.429 |
| In-geo | All Lengths | 29.464 | 176.786 | 29.464 | 2.946 | 216.071 | |
| Gemini 3.1 Flash Lite | Global | All Lengths | 4.464 | 26.786 | 4.464 | 0.446 | 89.286 |
| In-geo | All Lengths | 4.911 | 29.464 | 4.911 | 0.491 | 98.214 | |
| Gemini 3.0 / 3.1 Pro | Global/In-geo | Short Context | 35.714 | 214.286 | 35.714 | 3.571 | 230.429 |
| Long Context (>200k tokens) | 71.429 | 321.429 | 71.429 | 7.143 | 230.429 | ||
| Gemini 3.0 Flash | Global/In-geo | All Lengths | 8.929 | 53.571 | 8.929 | 0.893 | 125.000 |
| Gemini 2.5 Pro | Global/In-geo | Short Context | 22.321 | 178.571 | 22.321 | 2.232 | 164.286 |
| Long Context (>200k tokens) | 44.643 | 267.857 | 44.643 | 4.464 | 164.286 | ||
| Gemini 2.5 Flash | Global/In-geo | All Lengths | 5.357 | 44.643 | 5.357 | 0.536 | 107.143 |
| Gemini 2.5 Flash Lite | Global/In-geo | All Lengths | 1.786 | 7.143 | 1.786 | 0.179 | n/a |
NOTE: The Gemini model DBU rates shown here do not include a promotional discount of 20% (promotional pricing is 20% lower than shown). The promotion will run until June 30, 2026 after which all prices will revert to the DBU rates shown in this table.
Paga por uso con una prueba gratuita de 14 días o contáctanos para obtener descuentos por compromiso de uso o para requisitos personalizados.