Skip to content

Commit fad07e1

Browse files
foundry-samples-repo-sync[bot]sanjeev3Copilot
authored
Automated sync from private repo (2026-04-28) (#672)
* feat(model-router): add deployment scripts for routing modes and model subsets (#181) Co-authored-by: Copilot <copilot@github.com> * fixup: correct drift after commit replay * Update sync state to f5499aa --------- Co-authored-by: Sanjeev Jagtap <sajagtap@microsoft.com> Co-authored-by: Copilot <copilot@github.com> Co-authored-by: foundry-samples-repo-sync[bot] <foundry-samples-repo-sync[bot]@users.noreply.github.com>
1 parent 455710a commit fad07e1

6 files changed

Lines changed: 256 additions & 1 deletion

File tree

‎.gitattributes‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
*.sh text eol=lf

‎.github/.sync-sha‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
ebe7f93a317cd310a129ffb1fbb41e64e296d793
1+
f5499aad2868c10dac610a1afcf51809ba99a973

‎samples/REST/model-router/deploy-model-router-all-configs.sh‎

Lines changed: 82 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,82 @@
1+
# Deploys a model-router deployment with both a routing mode and a custom model subset.
2+
# WSL example: bash /mnt/c/Work/repos/foundry-samples-pr/samples/REST/model-router/deploy-model-router-all-configs.sh quality
3+
4+
SUBSCRIPTION_ID="<subscription-id>"
5+
RESOURCE_GROUP="<resource-group>"
6+
ACCOUNT_NAME="<azure-ai-foundry-account-name>"
7+
DEPLOYMENT_NAME="<deployment-name>"
8+
API_VERSION="2025-10-01-preview"
9+
SKU_NAME="GlobalStandard"
10+
SKU_CAPACITY="10"
11+
MODEL_ROUTER_FORMAT="OpenAI"
12+
MODEL_ROUTER_NAME="model-router"
13+
MODEL_ROUTER_VERSION="2025-11-18"
14+
# Valid values: balanced (default) | cost | quality
15+
ROUTING_MODE="balanced"
16+
SUBSET_1_FORMAT="OpenAI"
17+
SUBSET_1_NAME="gpt-4.1"
18+
SUBSET_1_VERSION="2025-04-14"
19+
SUBSET_2_FORMAT="OpenAI"
20+
SUBSET_2_NAME="gpt-5.2-chat"
21+
SUBSET_2_VERSION="2025-12-11"
22+
SUBSET_3_FORMAT="Meta"
23+
SUBSET_3_NAME="Llama-4-Maverick-17B-128E-Instruct-FP8"
24+
SUBSET_3_VERSION="1"
25+
26+
if [ $# -gt 0 ]; then
27+
ROUTING_MODE="$1"
28+
fi
29+
30+
if [ -z "${AZURE_AI_AUTH_TOKEN:-}" ]; then
31+
AZURE_AI_AUTH_TOKEN="$(az account get-access-token --resource https://management.azure.com --query accessToken -o tsv)"
32+
fi
33+
34+
case "${ROUTING_MODE}" in
35+
balanced|cost|quality)
36+
;;
37+
*)
38+
echo "Invalid ROUTING_MODE: ${ROUTING_MODE}. Valid values: balanced | cost | quality" >&2
39+
exit 1
40+
;;
41+
esac
42+
43+
# <deploy_model_router_all_configs>
44+
curl -X PUT "https://management.azure.com/subscriptions/${SUBSCRIPTION_ID}/resourceGroups/${RESOURCE_GROUP}/providers/Microsoft.CognitiveServices/accounts/${ACCOUNT_NAME}/deployments/${DEPLOYMENT_NAME}?api-version=${API_VERSION}" \
45+
-H "Content-Type: application/json" \
46+
-H "Authorization: Bearer $AZURE_AI_AUTH_TOKEN" \
47+
-d @- <<EOF
48+
{
49+
"sku": {
50+
"name": "${SKU_NAME}",
51+
"capacity": ${SKU_CAPACITY}
52+
},
53+
"properties": {
54+
"model": {
55+
"format": "${MODEL_ROUTER_FORMAT}",
56+
"name": "${MODEL_ROUTER_NAME}",
57+
"version": "${MODEL_ROUTER_VERSION}"
58+
},
59+
"routing": {
60+
"mode": "${ROUTING_MODE}",
61+
"models": [
62+
{
63+
"format": "${SUBSET_1_FORMAT}",
64+
"name": "${SUBSET_1_NAME}",
65+
"version": "${SUBSET_1_VERSION}"
66+
},
67+
{
68+
"format": "${SUBSET_2_FORMAT}",
69+
"name": "${SUBSET_2_NAME}",
70+
"version": "${SUBSET_2_VERSION}"
71+
},
72+
{
73+
"format": "${SUBSET_3_FORMAT}",
74+
"name": "${SUBSET_3_NAME}",
75+
"version": "${SUBSET_3_VERSION}"
76+
}
77+
]
78+
}
79+
}
80+
}
81+
EOF
82+
# </deploy_model_router_all_configs>

‎samples/REST/model-router/deploy-model-router-model-subset.sh‎

Lines changed: 66 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,66 @@
1+
# Deploys a model-router deployment that routes only to a custom subset of models.
2+
# WSL example: bash /mnt/c/Work/repos/foundry-samples-pr/samples/REST/model-router/deploy-model-router-model-subset.sh
3+
4+
SUBSCRIPTION_ID="<subscription-id>"
5+
RESOURCE_GROUP="<resource-group>"
6+
ACCOUNT_NAME="<azure-ai-foundry-account-name>"
7+
DEPLOYMENT_NAME="<deployment-name>"
8+
API_VERSION="2025-10-01-preview"
9+
SKU_NAME="GlobalStandard"
10+
SKU_CAPACITY="10"
11+
MODEL_ROUTER_FORMAT="OpenAI"
12+
MODEL_ROUTER_NAME="model-router"
13+
MODEL_ROUTER_VERSION="2025-11-18"
14+
SUBSET_1_FORMAT="OpenAI"
15+
SUBSET_1_NAME="gpt-4.1"
16+
SUBSET_1_VERSION="2025-04-14"
17+
SUBSET_2_FORMAT="OpenAI"
18+
SUBSET_2_NAME="gpt-5.2-chat"
19+
SUBSET_2_VERSION="2025-12-11"
20+
SUBSET_3_FORMAT="Meta"
21+
SUBSET_3_NAME="Llama-4-Maverick-17B-128E-Instruct-FP8"
22+
SUBSET_3_VERSION="1"
23+
24+
if [ -z "${AZURE_AI_AUTH_TOKEN:-}" ]; then
25+
AZURE_AI_AUTH_TOKEN="$(az account get-access-token --resource https://management.azure.com --query accessToken -o tsv)"
26+
fi
27+
28+
# <deploy_model_router_model_subset>
29+
curl -X PUT "https://management.azure.com/subscriptions/${SUBSCRIPTION_ID}/resourceGroups/${RESOURCE_GROUP}/providers/Microsoft.CognitiveServices/accounts/${ACCOUNT_NAME}/deployments/${DEPLOYMENT_NAME}?api-version=${API_VERSION}" \
30+
-H "Content-Type: application/json" \
31+
-H "Authorization: Bearer $AZURE_AI_AUTH_TOKEN" \
32+
-d @- <<EOF
33+
{
34+
"sku": {
35+
"name": "${SKU_NAME}",
36+
"capacity": ${SKU_CAPACITY}
37+
},
38+
"properties": {
39+
"model": {
40+
"format": "${MODEL_ROUTER_FORMAT}",
41+
"name": "${MODEL_ROUTER_NAME}",
42+
"version": "${MODEL_ROUTER_VERSION}"
43+
},
44+
"routing": {
45+
"models": [
46+
{
47+
"format": "${SUBSET_1_FORMAT}",
48+
"name": "${SUBSET_1_NAME}",
49+
"version": "${SUBSET_1_VERSION}"
50+
},
51+
{
52+
"format": "${SUBSET_2_FORMAT}",
53+
"name": "${SUBSET_2_NAME}",
54+
"version": "${SUBSET_2_VERSION}"
55+
},
56+
{
57+
"format": "${SUBSET_3_FORMAT}",
58+
"name": "${SUBSET_3_NAME}",
59+
"version": "${SUBSET_3_VERSION}"
60+
}
61+
]
62+
}
63+
}
64+
}
65+
EOF
66+
# </deploy_model_router_model_subset>

‎samples/REST/model-router/deploy-model-router-routing-mode.sh‎

Lines changed: 57 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,57 @@
1+
2+
# Deploys a model-router deployment with a selected routing mode.
3+
# WSL example: bash /mnt/c/Work/repos/foundry-samples-pr/samples/REST/model-router/deploy-model-router-routing-mode.sh quality
4+
5+
SUBSCRIPTION_ID="<subscription-id>"
6+
RESOURCE_GROUP="<resource-group>"
7+
ACCOUNT_NAME="<azure-ai-foundry-account-name>"
8+
DEPLOYMENT_NAME="<deployment-name>"
9+
API_VERSION="2025-10-01-preview"
10+
# Valid values: balanced (default) | cost | quality
11+
ROUTING_MODE="balanced"
12+
SKU_NAME="GlobalStandard"
13+
SKU_CAPACITY="10"
14+
MODEL_FORMAT="OpenAI"
15+
MODEL_NAME="model-router"
16+
MODEL_VERSION="2025-11-18"
17+
18+
if [ $# -gt 0 ]; then
19+
ROUTING_MODE="$1"
20+
fi
21+
22+
if [ -z "${AZURE_AI_AUTH_TOKEN:-}" ]; then
23+
AZURE_AI_AUTH_TOKEN="$(az account get-access-token --resource https://management.azure.com --query accessToken -o tsv)"
24+
fi
25+
26+
case "${ROUTING_MODE}" in
27+
balanced|cost|quality)
28+
;;
29+
*)
30+
echo "Invalid ROUTING_MODE: ${ROUTING_MODE}. Valid values: balanced | cost | quality" >&2
31+
exit 1
32+
;;
33+
esac
34+
35+
# <deploy_model_router_routing_mode>
36+
curl -X PUT "https://management.azure.com/subscriptions/${SUBSCRIPTION_ID}/resourceGroups/${RESOURCE_GROUP}/providers/Microsoft.CognitiveServices/accounts/${ACCOUNT_NAME}/deployments/${DEPLOYMENT_NAME}?api-version=${API_VERSION}" \
37+
-H "Content-Type: application/json" \
38+
-H "Authorization: Bearer $AZURE_AI_AUTH_TOKEN" \
39+
-d @- <<EOF
40+
{
41+
"sku": {
42+
"name": "${SKU_NAME}",
43+
"capacity": ${SKU_CAPACITY}
44+
},
45+
"properties": {
46+
"model": {
47+
"format": "${MODEL_FORMAT}",
48+
"name": "${MODEL_NAME}",
49+
"version": "${MODEL_VERSION}"
50+
},
51+
"routing": {
52+
"mode": "${ROUTING_MODE}"
53+
}
54+
}
55+
}
56+
EOF
57+
# </deploy_model_router_routing_mode>

‎‎

Lines changed: 49 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,49 @@
1+
2+
3+
SUBSCRIPTION_ID="<subscription-id>"
4+
RESOURCE_GROUP="<resource-group>"
5+
ACCOUNT_NAME="<azure-ai-foundry-account-name>"
6+
DEPLOYMENT_NAME="<deployment-name>"
7+
API_VERSION="2025-10-01-preview"
8+
# Valid values: balanced (default) | cost | quality
9+
ROUTING_MODE="${ROUTING_MODE:-balanced}"
10+
SKU_NAME="GlobalStandard"
11+
SKU_CAPACITY="10"
12+
MODEL_FORMAT="OpenAI"
13+
MODEL_NAME="model-router"
14+
MODEL_VERSION="2025-11-18"
15+
16+
if [ -z "${AZURE_AI_AUTH_TOKEN:-}" ]; then
17+
AZURE_AI_AUTH_TOKEN="$(az account get-access-token --resource https://management.azure.com --query accessToken -o tsv)"
18+
fi
19+
20+
case "${ROUTING_MODE}" in
21+
balanced|cost|quality)
22+
;;
23+
*)
24+
echo "Invalid ROUTING_MODE: ${ROUTING_MODE}. Valid values: balanced | cost | quality" >&2
25+
exit 1
26+
;;
27+
esac
28+
29+
curl -X PUT "https://management.azure.com/subscriptions/${SUBSCRIPTION_ID}/resourceGroups/${RESOURCE_GROUP}/providers/Microsoft.CognitiveServices/accounts/${ACCOUNT_NAME}/deployments/${DEPLOYMENT_NAME}?api-version=${API_VERSION}" \
30+
-H "Content-Type: application/json" \
31+
-H "Authorization: Bearer $AZURE_AI_AUTH_TOKEN" \
32+
-d @- <<EOF
33+
{
34+
"sku": {
35+
"name": "${SKU_NAME}",
36+
"capacity": ${SKU_CAPACITY}
37+
},
38+
"properties": {
39+
"model": {
40+
"format": "${MODEL_FORMAT}",
41+
"name": "${MODEL_NAME}",
42+
"version": "${MODEL_VERSION}"
43+
},
44+
"routing": {
45+
"mode": "${ROUTING_MODE}"
46+
}
47+
}
48+
}
49+
EOF

0 commit comments

Comments
 (0)