Proprietary Foundation Model Serving
Serve state-of-the-art proprietary foundation models for real-time and batch inference workload needs. This enables you to quickly and easily build applications that leverage high-quality proprietary generative AI models from various vendors directly on the Databricks platform without the need to additionally and separately engage with other vendors.
* For Azure customers, if you have an Azure Commit with Databricks, Databricks may make available this service as an ADI Service that integrates with Azure Databricks. The ADI Service is sold and invoiced by Databricks. Contact Sales to get access.
1. Azure Databricks, as a first-party service on Microsoft Azure, offers unified billing and support by Microsoft
The Premium tier on Azure Databricks corresponds to the Enterprise tier on AWS and GCP
Proprietary Foundation Model Serving DBU rates
Pay Per Token
Model | Standard Pay Per Token (DBU Per 1M Tokens) | ||||
|---|---|---|---|---|---|
Input | Output | Cache read | Cache write | ||
| GPT-5.6 Sol* ⌖ | Short context | 57.143 | 285.714 | 5.714 | 71.429 |
| Long context | 114.286 | 428.571 | 11.429 | 142.857 | |
| GPT-5.6 Terra ⌖ | Short context | 28.571 | 171.429 | 2.857 | 35.714 |
| Long context | 57.143 | 257.143 | 5.714 | 71.429 | |
| GPT-5.6 Luna ⌖ | Short context | 2.857 | 17.143 | 0.286 | 3.571 |
| Long context | 5.714 | 25.714 | 0.571 | 7.143 | |
| GPT-5.4 Pro, 5.5 Pro ⌖ | Short context | 428.571 | 2,571.43 | - | - |
| Long context | 857.143 | 3,857.14 | - | - | |
| GPT-5.5 ⌖ | Short context | 71.429 | 428.571 | 7.143 | - |
| Long context | 142.857 | 642.857 | 14.286 | - | |
| GPT-5.4 ⌖ | Short context | 35.714 | 214.286 | 3.571 | - |
| Long context | 71.429 | 321.429 | 7.143 | - | |
| GPT-5.4 Mini ⌖ | 10.714 | 64.286 | 1.071 | - | |
| GPT-5.4 Nano ⌖ | 2.857 | 17.857 | 0.286 | - | |
| GPT-5.2 Codex, 5.3 Codex ⌖ | 25 | 200 | 2.5 | - | |
| GPT-5.2 ⌖ | 25 | 200 | 2.5 | - | |
| GPT-5, 5.1 ⌖ | 17.857 | 142.857 | 1.786 | - | |
| GPT-5.1 Codex Max ⌖ | 17.857 | 142.857 | 1.786 | - | |
| GPT-5.1 Codex Mini ⌖ | 3.571 | 28.571 | 0.357 | - | |
| GPT-5 Mini ⌖ | 3.571 | 28.571 | 0.357 | - | |
| GPT-5 Nano ⌖ | 0.714 | 5.714 | 0.071 | - | |
| GPT Image 2 | Text tokens | 71.429 | - | 17.857 | - |
| Image tokens | 114.286 | 428.571 | 28.571 | - | |
| GPT Image 1.5 | Text tokens | 71.429 | 142.857 | 17.857 | - |
| Image tokens | 114.286 | 457.143 | 28.571 | - | |
| GPT Image 1 | Text tokens | 71.429 | - | 17.857 | - |
| Image tokens | 142.857 | 571.429 | 35.714 | - | |
| GPT Image 1 Mini | Text tokens | 28.571 | - | 2.857 | - |
| Image tokens | 35.714 | 114.286 | 3.571 | - | |
Priority Pay Per Token
Model | Priority Pay Per Token (DBU Per 1M Tokens) | ||||
|---|---|---|---|---|---|
Input | Output | Cache read | Cache write | ||
| GPT-5.6 Sol* ⌖ | Short context | 114.286 | 571.429 | 11.429 | 142.857 |
| Long context | 228.571 | 857.143 | 22.857 | 285.714 | |
| GPT-5.6 Terra ⌖ | Short context | 57.143 | 342.857 | 5.714 | 71.429 |
| Long context | 114.286 | 514.286 | 11.429 | 142.857 | |
| GPT-5.6 Luna ⌖ | Short context | 5.714 | 34.286 | 0.571 | 7.143 |
| Long context | 11.429 | 51.429 | 1.143 | 14.286 | |
| GPT-5.5 ⌖ | 178.571 | 1,071.43 | 17.857 | - | |
| GPT-5.4 ⌖ | 71.429 | 428.571 | 7.143 | - | |
| GPT-5.4 Mini ⌖ | 21.429 | 128.571 | 2.143 | - | |
| GPT-5.3 Codex ⌖ | 50 | 400 | 5 | - | |
| GPT-5.2 ⌖ | 50 | 400 | 5 | - | |
| GPT-5, 5.1 ⌖ | 35.714 | 285.714 | 3.571 | - | |
| GPT-5 Mini ⌖ | 6.429 | 51.429 | 0.643 | - | |
Batch Inference
Model | Batch Inference (DBU Per Hour) |
|---|---|
| GPT-5.6 Sol* ⌖ | 171.429 |
| GPT-5.6 Terra ⌖ | 153.571 |
| GPT-5.6 Luna ⌖ | 22.857 |
| GPT-5.4 Pro, 5.5 Pro ⌖ | 1,142.86 |
| GPT-5.5 ⌖ | 214.286 |
| GPT-5.4 ⌖ | 192.857 |
| GPT-5.4 Mini ⌖ | 107.143 |
| GPT-5.4 Nano ⌖ | 71.429 |
| GPT-5.2 ⌖ | 184.286 |
| GPT-5, 5.1 ⌖ | 131.429 |
| GPT-5 Mini ⌖ | 71.429 |
| GPT-5 Nano ⌖ | 53.571 |
⌖ When regional processing (data residency) is enabled, models with the ⌖ label will have a 10% uplift applied (will be charged DBU rates 10% greater than those shown in the table above).
Token modalities (text, image, audio) are shown explicitly only when their numeric prices differ. Once modalities are shown, separate rows also preserve differences in supported token types; a dash means that token type is not supported for that modality. Where modalities are not shown, each displayed price applies to every modality that supports the corresponding token type.
* NOTE: The GPT-5.6 Sol rates shown here reflect OpenAI’s promotional pricing, in effect through November 21, 2026. Afterward, the standard GPT-5.6 Sol rates will take effect; input, cache, and Batch rates will be 25% higher than those shown, while output rates will be 50% higher.
Proprietary Foundation Model Serving DBU rates
Pay Per Token
Model | Standard Pay Per Token (DBU Per 1M Tokens) | ||||
|---|---|---|---|---|---|
Input | Output | Cache read | Cache write | Cache write (1hr) | |
| Claude Fable 5.1 ⌖ | 142.857 | 714.286 | 3.571 | 178.571 | 285.714 |
| Claude Fable 5 ⌖ | 142.857 | 714.286 | 14.286 | 178.571 | 285.714 |
| Claude Opus 4.5, 4.6, 4.7, 4.8, 5 ⌖ | 71.429 | 357.143 | 7.143 | 89.286 | 142.857 |
| Claude Opus 4, 4.1 | 214.286 | 1,071.43 | 21.429 | 267.857 | 428.571 |
| Claude Sonnet 5 ⌖ | 28.571 | 142.857 | 2.857 | 35.714 | 57.143 |
| Claude Sonnet 4.5, 4.6 ⌖ | 42.857 | 214.286 | 4.286 | 53.571 | 85.714 |
| Claude Sonnet 4 | 42.857 | 214.286 | 4.286 | 53.571 | - |
| Claude Haiku 4.5 ⌖ | 14.286 | 71.429 | 1.429 | 17.857 | 28.571 |
Batch Inference
Model | Batch Inference (DBU Per Hour) |
|---|---|
| Claude Fable 5, 5.1 ⌖ | 357.143 |
| Claude Opus 4.5, 4.6, 4.7, 4.8, 5 ⌖ | 178.571 |
| Claude Opus 4, 4.1 | 514.286 |
| Claude Sonnet 5 ⌖ | 142.857 |
| Claude Sonnet 4.5, 4.6 ⌖ | 214.286 |
| Claude Sonnet 4 | 214.286 |
| Claude Haiku 4.5 ⌖ | 114.286 |
⌖ When regional processing (data residency) is enabled, models with the ⌖ label will have a 10% uplift applied (will be charged DBU rates 10% greater than those shown in the table above).
Proprietary Foundation Model Serving DBU rates
Pay Per Token
Model |
| Standard Pay Per Token (DBU Per 1M Tokens) | ||
|---|---|---|---|---|
Input | Output | Cache read | ||
| Gemini 3.0 Pro, 3.1 Pro* | Short context | 35.714 | 214.286 | 3.571 |
| Long context | 71.429 | 321.429 | 7.143 | |
| Gemini 2.5 Pro* | Short context | 22.321 | 178.571 | 2.232 |
| Long context | 44.643 | 267.857 | 4.464 | |
| Gemini 3.7 Flash, 3.8 Flash** ⌖ | 10.714 | 53.571 | 1.071 | |
| Gemini 3.6 Flash** | 10.714 | 53.571 | 1.071 | |
| Gemini 3.5 Flash* ⌖ | 26.786 | 160.714 | 2.679 | |
| Gemini 3.0 Flash* | Text tokens | 8.929 | 53.571 | 0.893 |
| Image tokens | 8.929 | - | 0.893 | |
| Audio tokens | 17.857 | - | 1.786 | |
| Gemini 2.5 Flash* | Text tokens | 5.357 | 44.643 | 0.536 |
| Image tokens | 5.357 | - | 0.536 | |
| Audio tokens | 17.857 | - | 1.786 | |
| Gemini 3.5 Flash Lite* ⌖ | 5.357 | 44.643 | 0.536 | |
| Gemini 3.1 Flash Lite* ⌖ | Text tokens | 4.464 | 26.786 | 0.446 |
| Image tokens | 4.464 | - | 0.446 | |
| Audio tokens | 8.929 | - | 0.893 | |
| Gemini 2.5 Flash Lite* | Text tokens | 1.786 | 7.143 | 0.179 |
| Image tokens | 1.786 | - | 0.179 | |
| Audio tokens | 5.357 | - | 0.536 | |
| Gemini 3 Pro Image* ⌖ | Text tokens | 35.714 | 214.286 | - |
| Image tokens | 35.714 | 2,142.86 | - | |
| Gemini 3.1 Flash Image* ⌖ | Text tokens | 8.929 | 53.571 | - |
| Image tokens | 8.929 | 1,071.43 | - | |
| Gemini 3.1 Flash Lite Image* | Text tokens | 4.464 | 26.786 | - |
| Image tokens | 4.464 | 535.714 | - | |
Priority Pay Per Token
Model |
| Priority Pay Per Token (DBU Per 1M Tokens) | ||
|---|---|---|---|---|
Input | Output | Cache read | ||
| Gemini 3.1 Pro* | Short context | 64.286 | 385.714 | 6.429 |
| Long context | 128.571 | 578.571 | 12.857 | |
| Gemini 3.7 Flash, 3.8 Flash** ⌖ | 19.286 | 96.429 | 1.929 | |
| Gemini 3.6 Flash** | 19.286 | 96.429 | 1.929 | |
| Gemini 3.5 Flash* ⌖ | 48.214 | 289.286 | 4.821 | |
| Gemini 3.0 Flash* | Text tokens | 16.071 | 96.429 | 1.607 |
| Image tokens | 16.071 | - | 1.607 | |
| Audio tokens | 32.143 | - | 3.214 | |
| Gemini 3.5 Flash Lite* ⌖ | 9.643 | 80.357 | 0.964 | |
| Gemini 3.1 Flash Lite* ⌖ | Text tokens | 8.036 | 48.214 | 0.804 |
| Image tokens | 8.036 | - | 0.804 | |
| Audio tokens | 16.071 | - | 1.607 | |
Batch Inference
Model | Batch Inference (DBU Per Hour) |
|---|---|
| Gemini 3.0 Pro, 3.1 Pro* | 230.429 |
| Gemini 2.5 Pro* | 164.286 |
| Gemini 3.7 Flash, 3.8 Flash** ⌖ | 74.286 |
| Gemini 3.6 Flash** | 74.286 |
| Gemini 3.5 Flash* ⌖ | 196.429 |
| Gemini 3.0 Flash* | 125 |
| Gemini 2.5 Flash* | 107.143 |
| Gemini 3.5 Flash Lite* ⌖ | 121.429 |
| Gemini 3.1 Flash Lite* ⌖ | 89.286 |
⌖ When regional processing (data residency) is enabled, models with the ⌖ label will have a 10% uplift applied (will be charged DBU rates 10% greater than those shown in the table above).
Token modalities (text, image, audio) are shown explicitly only when their numeric prices differ. Once modalities are shown, separate rows also preserve differences in supported token types; a dash means that token type is not supported for that modality. Where modalities are not shown, each displayed price applies to every modality that supports the corresponding token type.
* The DBU rates for these Gemini models do not include a promotional discount of 20% (promotional pricing is 20% lower than shown). The promotion expires on Jan 31, 2027, after which date all prices will revert to the DBU rates shown in this table.
** The Gemini 3.6 Flash, 3.7 Flash, and 3.8 Flash DBU rates shown here reflect a 50% promotion in effect through December 31, 2026. Afterward, the standard rates, twice those shown, will take effect.
Proprietary Foundation Model Serving DBU rates
Pay Per Token
Model | Standard Pay Per Token (DBU Per 1M Tokens) | ||
|---|---|---|---|
Input | Output | Cache read | |
| Grok 4.6* ⌖ | 35.714 | 107.143 | 8.929 |
Batch Inference
Model | Batch Inference (DBU Per Hour) |
|---|---|
| Grok 4.6* ⌖ | 153.571 |
⌖ When regional processing (data residency) is enabled, models with the ⌖ label will have a 10% uplift applied (will be charged DBU rates 10% greater than those shown in the table above).
* The DBU rates for these Grok models do not include a promotional discount of 20% (promotional pricing is 20% lower than shown). The promotion expires on Jan 31, 2027, after which date all prices will revert to the DBU rates shown in this table.
Pay as you go with a 14-day free trial or contact us for committed-use discounts or custom requirements.