You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: articles/ai-services/openai/how-to/provisioned-throughput-onboarding.md
+6-6Lines changed: 6 additions & 6 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -76,7 +76,7 @@ From Azure AI Studio select **Quota** in the left hand navigation bar, and then
76
76
77
77
:::image type="content" source="../media/provisioned/quota-alternate.png" alt-text="Screenshot of new quota UI for Azure OpenAI provisioned." lightbox="../media/provisioned/quota-alternate.png":::
78
78
79
-
Here, you can view the quota granted in the selected subscription/region, and how much is used. For example, the screenshot above shows that 100 out of 200 PTUs of quota are used in South Central US. This means that 100 are available for use in creating new deployments.
79
+
Here, you can view the quota granted in the selected subscription/region, and how much is used. For example, the screenshot above shows that 100 out of 200 PTUs of quota are used in South Central US. This means that 100 are available for use in creating new deployments.**MICHAEL: WILL NEED TO UPDATE SCREENSHOTS TO FINAL VERSIONS AND ALSO EDIT THE PTUS IN THIS TEXT**
80
80
81
81
By selecting the small arrow to the left of the quota name **Provisioned Managed Throughput Unit**, you can expand it to show the deployments contributing to the usage. In the example above, the “gpt-4-chatbot" deployment in the “Production-Deployments” resource is the deployment using 100 PTUs of the quota.
82
82
@@ -93,7 +93,7 @@ Provisioned Throughput deployments are created via Azure OpenAI resource objects
93
93
94
94
## Create provisioned throughput deployments
95
95
96
-
1. Launch [Azure AI Studio](https://ai.azure.com/)
96
+
1. Launch [Azure AI Studio](https://ai.azure.com/)**MICHAEL: Let's update all Studio references/screenshots to be Azure OpenAI studio with the experience that'll be default on 8/12. I've verifying the plan with Shane**
97
97
2. Select the Azure OpenAI resource in the desired region, then select **Deployments** in the left-hand navigation bar.
98
98
99
99
:::image type="content" source="../media/provisioned/deployments.png" alt-text="Screenshot of deployments UI for Azure OpenAI deployments." lightbox="../media/provisioned/deployments.png":::
@@ -107,7 +107,7 @@ Provisioned Throughput deployments are created via Azure OpenAI resource objects
107
107
:::image type="content" source="../media/provisioned/deploy-model.png" alt-text="Screenshot of model deployment UI for Azure OpenAI with a provisioned model being deployed." lightbox="../media/provisioned/deploy-model.png":::
108
108
109
109
- Provide the deployment name and model version.
110
-
- Specify the Deployment Type as **ProvisionedManaged**. This is what indicates a provisioned, instead of a standard deployment.
110
+
- Specify the Deployment Type as **Provisioned-Managed**. This is what indicates a provisioned, instead of a standard deployment.
111
111
- Note the message reminding you to purchase an Azure Reservation to obtain a discount for a term commitment.
112
112
113
113
5. Evaluate capacity availability.
@@ -124,11 +124,11 @@ Provisioned Throughput deployments are created via Azure OpenAI resource objects
124
124
125
125
7. If there's insufficient capacity, choose another region.
126
126
127
-
If you select more PTUs than are available as service capacity, you're given the option to choose a resource in a region that has available quota and capacity.
127
+
If you select more PTUs than are available as service capacity, you're given the option to choose a resource in a different region that may have more quota and capacity.
128
128
129
129
:::image type="content" source="../media/provisioned/insufficient-capacity.png" alt-text="Screenshot of capacity recommendation to try a new region." lightbox="../media/provisioned/insufficient-capacity.png":::
130
130
131
-
Selecting **See other regions** brings up a dialog where you can select alternative regions with both available quota and capacity for this model. Select a new resource and the deployment dialog will redisplay with the new resource so that you can continue your deployment.
131
+
Selecting **See other regions** brings up a dialog where you can select alternative regions where you can create a deployment of the selected model and version. The dialog shows you the maximum sized deployment that you can create in the region based on both quota availability and service capacity availability. Select a new resource and the deployment dialog will redisplay with the new resource so that you can continue your deployment.
132
132
133
133
:::image type="content" source="../media/provisioned/different-region.png" alt-text="Screenshot of choose a different region with more capacity option UI." lightbox="../media/provisioned/different-region.png":::
134
134
@@ -158,7 +158,7 @@ To buy a reservation:
158
158
159
159
:::image type="content" source="../media/provisioned/reservation-pane.png" alt-text="Screenshot of the add reservations portal experience." lightbox="../media/provisioned/reservation-pane.png":::
160
160
161
-
3. Select **Azure OpenAI** from the reservation catalog.
161
+
3. Select **Azure OpenAI Service Provisioned** from the reservation catalog.**MICHAEL: WE'LL NEED A NEW SCREENSHOT**
162
162
163
163
:::image type="content" source="../media/provisioned/purchase.png" alt-text="Screenshot of the purchase reservation experience." lightbox="../media/provisioned/purchase.png":::
0 commit comments