diff --git a/.github/workflows/strix.yml b/.github/workflows/strix.yml
index c8c054e422..9b30488388 100644
--- a/.github/workflows/strix.yml
+++ b/.github/workflows/strix.yml
@@ -674,7 +674,7 @@ jobs:
           echo "LLM_API_BASE_FILE=$llm_api_base_file" >> "$GITHUB_ENV"
 
       - name: Prepare GitHub Models fallback credentials
-        if: steps.gate.outputs.provider_mode == 'openai_direct' || steps.gate.outputs.provider_mode == 'openrouter' || steps.gate.outputs.provider_mode == 'nvidia_nim'
+        if: steps.gate.outputs.provider_mode == 'github_models' || steps.gate.outputs.provider_mode == 'openai_direct' || steps.gate.outputs.provider_mode == 'openrouter' || steps.gate.outputs.provider_mode == 'nvidia_nim'
         env:
           GITHUB_MODELS_FALLBACK_TOKEN: ${{ secrets.STRIX_GITHUB_MODELS_TOKEN || github.token }}
           OPENAI_FALLBACK_KEY: ${{ secrets.STRIX_OPENAI_API_KEY || secrets.OPENAI_API_KEY }}
@@ -706,6 +706,9 @@ jobs:
             openai_fallback_key_file="$RUNNER_TEMP/openai_fallback_key.txt"
             printf '%s' "$openai_trimmed" > "$openai_fallback_key_file"
             echo "STRIX_OPENAI_FALLBACK_KEY_FILE=$openai_fallback_key_file" >> "$GITHUB_ENV"
+            openai_fallback_api_base_file="$RUNNER_TEMP/openai_fallback_api_base.txt"
+            printf '%s' 'https://api.openai.com/v1' > "$openai_fallback_api_base_file"
+            echo "STRIX_OPENAI_FALLBACK_API_BASE_FILE=$openai_fallback_api_base_file" >> "$GITHUB_ENV"
           fi
 
       - name: Prepare Vertex AI credentials
@@ -838,6 +841,7 @@ jobs:
           STRIX_GITHUB_MODELS_API_BASE_FILE: ${{ env.STRIX_GITHUB_MODELS_API_BASE_FILE }}
           STRIX_GITHUB_MODELS_KEY_FILE: ${{ env.STRIX_GITHUB_MODELS_KEY_FILE }}
           STRIX_OPENAI_FALLBACK_KEY_FILE: ${{ env.STRIX_OPENAI_FALLBACK_KEY_FILE }}
+          STRIX_OPENAI_FALLBACK_API_BASE_FILE: ${{ env.STRIX_OPENAI_FALLBACK_API_BASE_FILE }}
           STRIX_FAIL_ON_PROVIDER_SIGNAL: "1"
           STRIX_VERTEX_FALLBACK_MODELS: ""
           NPM_CONFIG_IGNORE_SCRIPTS: "true"
diff --git a/scripts/ci/strix_quick_gate.sh b/scripts/ci/strix_quick_gate.sh
index 36ec3e5f8e..fa676a0d5a 100755
--- a/scripts/ci/strix_quick_gate.sh
+++ b/scripts/ci/strix_quick_gate.sh
@@ -2417,7 +2417,14 @@ resolved_llm_api_base_for_model() {
 
 	local api_base_file="$LLM_API_BASE_FILE"
 	local api_base_file_name="LLM_API_BASE_FILE"
-	if is_github_models_model "$model" && [ -n "${STRIX_GITHUB_MODELS_API_BASE_FILE:-}" ]; then
+	if is_explicit_openai_model "$model" && [ -n "${STRIX_OPENAI_FALLBACK_API_BASE_FILE:-}" ]; then
+		# Cross-provider fallback: openai-direct/* candidates must reach the
+		# direct OpenAI API even when the primary provider selected a
+		# different LLM_API_BASE_FILE endpoint (e.g. NVIDIA NIM). Without
+		# this the fallback hits the primary gateway and 404s.
+		api_base_file="$STRIX_OPENAI_FALLBACK_API_BASE_FILE"
+		api_base_file_name="STRIX_OPENAI_FALLBACK_API_BASE_FILE"
+	elif is_github_models_model "$model" && [ -n "${STRIX_GITHUB_MODELS_API_BASE_FILE:-}" ]; then
 		# Cross-provider fallback: when the active primary provider uses a
 		# different API base (for example OpenRouter), github_models/* fallback
 		# attempts must still route through the GitHub Models inference endpoint.
diff --git a/scripts/ci/test_strix_quick_gate.sh b/scripts/ci/test_strix_quick_gate.sh
index 945eb3fb30..571bf4dd9a 100755
--- a/scripts/ci/test_strix_quick_gate.sh
+++ b/scripts/ci/test_strix_quick_gate.sh
@@ -365,6 +365,9 @@ assert_strix_workflow_pr_trigger_hardened() {
 	assert_file_contains "$workflow_file" "steps.gate.outputs.provider_mode == 'nvidia_nim' && 'nvidia_nim/nvidia/llama-3.3-nemotron-super-49b-v1.5 openai-direct/gpt-5.6-luna'" "strix workflow gives NVIDIA NIM scans contracted fallbacks"
 	assert_file_not_contains "$workflow_file" "STRIX_FALLBACK_MODELS: \${{ steps.gate.outputs.provider_mode == 'github_models' && 'github_models/openai/o3" "strix workflow fallback list must not depend on GitHub Models, which is in platform-wide retirement"
 	assert_file_contains "$workflow_file" "Prepare GitHub Models fallback credentials" "strix workflow provisions GitHub Models fallback credentials for direct-OpenAI scans"
+	assert_file_contains "$workflow_file" "steps.gate.outputs.provider_mode == 'github_models' || steps.gate.outputs.provider_mode == 'openai_direct'" "strix workflow provisions direct-OpenAI fallback credentials for GitHub Models scans"
+	assert_file_contains "$workflow_file" "STRIX_OPENAI_FALLBACK_API_BASE_FILE" "strix workflow routes direct-OpenAI fallbacks through a trusted API base file"
+	assert_file_contains "$workflow_file" "https://api.openai.com/v1" "strix workflow uses the OpenAI platform endpoint for direct fallbacks"
 	assert_file_contains "$GATE_SCRIPT" "STRIX_GITHUB_MODELS_KEY_FILE" "strix gate reads the optional GitHub Models fallback key file"
 	assert_file_contains "$GATE_SCRIPT" "STRIX_GITHUB_MODELS_API_BASE_FILE" "strix gate routes github_models fallback models through the GitHub Models endpoint"
 	assert_file_not_contains "$workflow_file" 'github_models/deepseek/deepseek-r1-0528 | github_models/deepseek/deepseek-v3-0324)' "strix workflow keeps DeepSeek GitHub Models restricted to fallback-only routing"
@@ -3421,6 +3424,54 @@ REPORT
 			;;
 		esac
 		;;
+	nvidia-nim-openai-direct-fallback-success)
+		case "${STRIX_LLM:-}" in
+		nvidia_nim/primary-model)
+			if [ "${LLM_API_KEY:-}" != "dummy" ] || [ "${LLM_API_BASE:-}" != "https://integrate.api.nvidia.com/v1" ]; then
+				echo "unexpected NVIDIA NIM primary credentials or endpoint" >&2
+				exit 17
+			fi
+			echo "nvidia_nim.RateLimitError: Error code: 429 - provider quota exhausted"
+			exit 1
+			;;
+		openai/gpt-5.6-luna)
+			if [ "${LLM_API_KEY:-}" != "openai-fallback-key" ] || [ "${LLM_API_BASE:-}" != "https://api.openai.com/v1" ]; then
+				echo "unexpected direct-OpenAI fallback credentials or endpoint" >&2
+				exit 18
+			fi
+			echo "scan ok after direct-OpenAI fallback"
+			exit 0
+			;;
+		*)
+			echo "unexpected NVIDIA-to-OpenAI fallback model ${STRIX_LLM:-}" >&2
+			exit 19
+			;;
+		esac
+		;;
+	github-models-openai-direct-fallback-success)
+		case "${STRIX_LLM:-}" in
+		openai/gpt-5)
+			if [ "${LLM_API_KEY:-}" != "dummy" ] || [ "${LLM_API_BASE:-}" != "https://models.github.ai/inference" ]; then
+				echo "unexpected GitHub Models primary credentials or endpoint" >&2
+				exit 20
+			fi
+			echo "openai.RateLimitError: Error code: 429 - GitHub Models quota exhausted"
+			exit 1
+			;;
+		openai/gpt-5.6-luna)
+			if [ "${LLM_API_KEY:-}" != "openai-fallback-key" ] || [ "${LLM_API_BASE:-}" != "https://api.openai.com/v1" ]; then
+				echo "unexpected GitHub-to-OpenAI fallback credentials or endpoint" >&2
+				exit 21
+			fi
+			echo "scan ok after GitHub-to-OpenAI fallback"
+			exit 0
+			;;
+		*)
+			echo "unexpected GitHub-to-OpenAI fallback model ${STRIX_LLM:-}" >&2
+			exit 22
+			;;
+		esac
+		;;
 	vertex-all-notfound)
 		echo "Error: litellm.NotFoundError: Vertex_aiException - x"
 		echo '"status": "NOT_FOUND"'
@@ -5648,6 +5699,18 @@ PY
 		env_cmd+=(STRIX_GITHUB_MODELS_API_BASE_FILE="$tmp_dir/github_models_api_base.txt")
 		env_cmd+=(STRIX_GITHUB_MODELS_KEY_FILE="$tmp_dir/github_models_key.txt")
 	fi
+	if [ "$scenario" = "nvidia-nim-openai-direct-fallback-success" ]; then
+		printf '%s' 'openai-fallback-key' >"$tmp_dir/openai_fallback_key.txt"
+		printf '%s' 'https://api.openai.com/v1' >"$tmp_dir/openai_fallback_api_base.txt"
+		env_cmd+=(STRIX_OPENAI_FALLBACK_KEY_FILE="$tmp_dir/openai_fallback_key.txt")
+		env_cmd+=(STRIX_OPENAI_FALLBACK_API_BASE_FILE="$tmp_dir/openai_fallback_api_base.txt")
+	fi
+	if [ "$scenario" = "github-models-openai-direct-fallback-success" ]; then
+		printf '%s' 'openai-fallback-key' >"$tmp_dir/openai_fallback_key.txt"
+		printf '%s' 'https://api.openai.com/v1' >"$tmp_dir/openai_fallback_api_base.txt"
+		env_cmd+=(STRIX_OPENAI_FALLBACK_KEY_FILE="$tmp_dir/openai_fallback_key.txt")
+		env_cmd+=(STRIX_OPENAI_FALLBACK_API_BASE_FILE="$tmp_dir/openai_fallback_api_base.txt")
+	fi
 	if [ "$min_fail_severity" = "__UNSET__" ]; then
 		local next_env_cmd=()
 		local env_pair
@@ -5902,6 +5965,68 @@ run_gate_case_allow_provider_signal() {
 	run_gate_case_with_provider_signal_mode "0" "$@"
 }
 
+run_nvidia_nim_openai_direct_fallback_case() {
+	run_gate_case_allow_provider_signal "nvidia-nim-openai-direct-fallback-success" \
+		"nvidia_nim/primary-model" \
+		"" \
+		"0" \
+		"REGEX:Strix quick scan succeeded with fallback model 'openai-direct/gpt-5.6-luna' in [0-9]+s\\." \
+		"2" \
+		"nvidia_nim/primary-model|openai/gpt-5.6-luna" \
+		"https://integrate.api.nvidia.com/v1|https://api.openai.com/v1" \
+		"nvidia_nim" \
+		"https://integrate.api.nvidia.com/v1" \
+		"" \
+		"0" \
+		"CRITICAL" \
+		"0" \
+		"" \
+		"" \
+		"1200" \
+		"0" \
+		"" \
+		"" \
+		"" \
+		"" \
+		"0" \
+		"" \
+		"" \
+		"" \
+		"__SAME_AS_FALLBACK_MODELS__" \
+		"openai-direct/gpt-5.6-luna"
+}
+
+run_github_models_openai_direct_fallback_case() {
+	run_gate_case_allow_provider_signal "github-models-openai-direct-fallback-success" \
+		"openai/gpt-5" \
+		"" \
+		"0" \
+		"REGEX:Strix quick scan succeeded with fallback model 'openai-direct/gpt-5.6-luna' in [0-9]+s\\." \
+		"2" \
+		"openai/gpt-5|openai/gpt-5.6-luna" \
+		"https://models.github.ai/inference|https://api.openai.com/v1" \
+		"openai" \
+		"https://models.github.ai/inference" \
+		"" \
+		"0" \
+		"CRITICAL" \
+		"0" \
+		"" \
+		"" \
+		"1200" \
+		"0" \
+		"" \
+		"" \
+		"" \
+		"" \
+		"0" \
+		"" \
+		"" \
+		"" \
+		"__SAME_AS_FALLBACK_MODELS__" \
+		"openai-direct/gpt-5.6-luna"
+}
+
 run_github_models_http410_case() {
 	local scenario="$1"
 	local expected_exit="$2"
@@ -6124,6 +6249,12 @@ run_filtered_gate_case_if_requested() {
 			"" \
 			"github_models/openai/o3"
 		;;
+	nvidia-nim-openai-direct-fallback-success)
+		run_nvidia_nim_openai_direct_fallback_case
+		;;
+	github-models-openai-direct-fallback-success)
+		run_github_models_openai_direct_fallback_case
+		;;
 	gemini-timeout-fallback-success)
 		run_gate_case_allow_provider_signal "gemini-timeout-fallback-success" \
 			"gemini/timeout-fallback-primary" \
@@ -12514,6 +12645,13 @@ run_gate_case "openai-direct-quota-github-models-fallback-success" \
 	"" \
 	"github_models/openai/o3"
 
+# NVIDIA NIM primary scans must route an explicit direct-OpenAI fallback to
+# the OpenAI endpoint and its own credential, not the primary provider input.
+run_nvidia_nim_openai_direct_fallback_case
+
+# GitHub Models primary scans use the same direct-OpenAI fallback contract.
+run_github_models_openai_direct_fallback_case
+
 run_gate_case "github-models-fallback-success-deepseek-v3" \
 	"vertex_ai/missing-primary" \
 	"github_models/deepseek/deepseek-r1-0528 github_models/deepseek/deepseek-v3-0324" \