diff --git a/.github/workflows/e2e-script.yaml b/.github/workflows/e2e-script.yaml
index a7c6bb6dee7..177e30ff954 100644
--- a/.github/workflows/e2e-script.yaml
+++ b/.github/workflows/e2e-script.yaml
@@ -229,8 +229,8 @@ jobs:
             printf 'NEMOCLAW_E2E_USE_HOSTED_INFERENCE=1\n'
             printf 'NEMOCLAW_PROVIDER=custom\n'
             printf 'NEMOCLAW_ENDPOINT_URL=https://inference-api.nvidia.com/v1\n'
-            printf 'NEMOCLAW_MODEL=nvidia/nemotron-3-ultra\n'
-            printf 'NEMOCLAW_COMPAT_MODEL=nvidia/nemotron-3-ultra\n'
+            printf 'NEMOCLAW_MODEL=nvidia/nvidia/nemotron-3-ultra\n'
+            printf 'NEMOCLAW_COMPAT_MODEL=nvidia/nvidia/nemotron-3-ultra\n'
             printf 'NEMOCLAW_PREFERRED_API=openai-completions\n'
             printf 'COMPATIBLE_API_KEY=%s\n' "${NVIDIA_INFERENCE_API_KEY}"
           } >> "$GITHUB_ENV"
diff --git a/.github/workflows/e2e-vitest-scenarios.yaml b/.github/workflows/e2e-vitest-scenarios.yaml
index 2ec40bd9539..04415d83873 100644
--- a/.github/workflows/e2e-vitest-scenarios.yaml
+++ b/.github/workflows/e2e-vitest-scenarios.yaml
@@ -1415,8 +1415,8 @@ jobs:
           NVIDIA_INFERENCE_API_KEY: ${{ secrets.NVIDIA_INFERENCE_API_KEY }}
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
         run: |
           set -euo pipefail
@@ -2135,8 +2135,8 @@ jobs:
       NEMOCLAW_NON_INTERACTIVE: "1"
       NEMOCLAW_PROVIDER: custom
       NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-      NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-      NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+      NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+      NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
       NEMOCLAW_PREFERRED_API: openai-completions
       NEMOCLAW_SANDBOX_NAME: e2e-rebuild-hermes
       OPENSHELL_GATEWAY: nemoclaw
@@ -2218,8 +2218,8 @@ jobs:
       NEMOCLAW_NON_INTERACTIVE: "1"
       NEMOCLAW_PROVIDER: custom
       NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-      NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-      NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+      NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+      NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
       NEMOCLAW_PREFERRED_API: openai-completions
       NEMOCLAW_SANDBOX_NAME: e2e-rebuild-hermes-base
       OPENSHELL_GATEWAY: nemoclaw
@@ -2970,8 +2970,8 @@ jobs:
       NEMOCLAW_SANDBOX_NAME: "e2e-resume"
       NEMOCLAW_PROVIDER: "custom"
       NEMOCLAW_ENDPOINT_URL: "https://inference-api.nvidia.com/v1"
-      NEMOCLAW_MODEL: "nvidia/nemotron-3-ultra"
-      NEMOCLAW_COMPAT_MODEL: "nvidia/nemotron-3-ultra"
+      NEMOCLAW_MODEL: "nvidia/nvidia/nemotron-3-ultra"
+      NEMOCLAW_COMPAT_MODEL: "nvidia/nvidia/nemotron-3-ultra"
     steps:
       - uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
         with:
diff --git a/.github/workflows/nightly-e2e.yaml b/.github/workflows/nightly-e2e.yaml
index a8893d37f57..b47644dff88 100644
--- a/.github/workflows/nightly-e2e.yaml
+++ b/.github/workflows/nightly-e2e.yaml
@@ -468,8 +468,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -544,8 +544,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_ISSUE_4434_LIVE: "1"
@@ -1003,8 +1003,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1301,8 +1301,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1608,8 +1608,8 @@ jobs:
           NVIDIA_INFERENCE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           E2E_ARTIFACT_DIR: ${{ github.workspace }}/e2e-artifacts/vitest/credential-migration
@@ -1834,8 +1834,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1847,8 +1847,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1886,8 +1886,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1899,8 +1899,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1938,8 +1938,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1951,8 +1951,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -1991,8 +1991,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -2004,8 +2004,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -2044,8 +2044,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -2058,8 +2058,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -2100,8 +2100,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -2114,8 +2114,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -2194,8 +2194,8 @@ jobs:
           NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
           NEMOCLAW_PROVIDER: custom
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
-          NEMOCLAW_MODEL: nvidia/nemotron-3-ultra
-          NEMOCLAW_COMPAT_MODEL: nvidia/nemotron-3-ultra
+          NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-ultra
+          NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-ultra
           NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
diff --git a/src/lib/inference/onboard-probes.test.ts b/src/lib/inference/onboard-probes.test.ts
index 555ccca6ea6..14e00c29c22 100644
--- a/src/lib/inference/onboard-probes.test.ts
+++ b/src/lib/inference/onboard-probes.test.ts
@@ -293,8 +293,8 @@ describe("OpenAI-compatible inference probes", () => {
   });
 
   it("bounds the hosted compatible inference probe for the served Nemotron model", () => {
-    expect(getChatCompletionsProbePayload("nvidia/nemotron-3-ultra")).toEqual({
-      model: "nvidia/nemotron-3-ultra",
+    expect(getChatCompletionsProbePayload("nvidia/nvidia/nemotron-3-ultra")).toEqual({
+      model: "nvidia/nvidia/nemotron-3-ultra",
       messages: [{ role: "user", content: "Reply with exactly: OK" }],
       max_tokens: 8,
     });
diff --git a/src/lib/onboard/providers.ts b/src/lib/onboard/providers.ts
index a91f435aa84..77a5eaceed0 100644
--- a/src/lib/onboard/providers.ts
+++ b/src/lib/onboard/providers.ts
@@ -27,7 +27,10 @@ const HOSTED_INFERENCE_SOURCE_ENV = "NVIDIA_INFERENCE_API_KEY";
 const HOSTED_INFERENCE_PROVIDER_KEY_ENV = "NEMOCLAW_PROVIDER_KEY";
 const HOSTED_INFERENCE_CREDENTIAL_ENV = "COMPATIBLE_API_KEY";
 const HOSTED_INFERENCE_ENDPOINT_URL = "https://inference-api.nvidia.com/v1";
-const HOSTED_INFERENCE_MODEL = "nvidia/nemotron-3-ultra";
+// Private CI-compatible Inference Hub endpoint model IDs use the
+// provider/namespace/model convention. This endpoint is staged as a custom
+// OpenAI-compatible provider, not as the public build.nvidia.com provider.
+const HOSTED_INFERENCE_MODEL = "nvidia/nvidia/nemotron-3-ultra";
 const NON_INTERACTIVE_PROVIDER_ALIASES = {
   cloud: "build",
   nim: "nim-local",
diff --git a/test/e2e-scenario/fixtures/hosted-inference.ts b/test/e2e-scenario/fixtures/hosted-inference.ts
index bb37f8858fa..4efcbbddbd5 100644
--- a/test/e2e-scenario/fixtures/hosted-inference.ts
+++ b/test/e2e-scenario/fixtures/hosted-inference.ts
@@ -6,7 +6,7 @@ const HOSTED_INFERENCE_CREDENTIAL_ENV = "COMPATIBLE_API_KEY";
 const HOSTED_INFERENCE_PROVIDER = "custom";
 const HOSTED_INFERENCE_PROVIDER_NAME = "compatible-endpoint";
 const DEFAULT_HOSTED_INFERENCE_BASE_URL = "https://inference-api.nvidia.com/v1";
-const DEFAULT_HOSTED_INFERENCE_MODEL = "nvidia/nemotron-3-ultra";
+const DEFAULT_HOSTED_INFERENCE_MODEL = "nvidia/nvidia/nemotron-3-ultra";
 
 export interface HostedInferenceSecrets {
   required(name: string): string;
diff --git a/test/e2e-scenario/live/agent-turn-latency-helpers.ts b/test/e2e-scenario/live/agent-turn-latency-helpers.ts
index 9f33e39d673..327ebdfdc44 100644
--- a/test/e2e-scenario/live/agent-turn-latency-helpers.ts
+++ b/test/e2e-scenario/live/agent-turn-latency-helpers.ts
@@ -23,7 +23,7 @@ export const HERMES_SANDBOX =
 validateSandboxName(OPENCLAW_SANDBOX);
 validateSandboxName(HERMES_SANDBOX);
 const DEFAULT_NVIDIA_MODEL = "nvidia/nemotron-3-super-120b-a12b";
-const DEFAULT_COMPAT_MODEL = "nvidia/nemotron-3-ultra";
+const DEFAULT_COMPAT_MODEL = "nvidia/nvidia/nemotron-3-ultra";
 const USE_COMPATIBLE_HOSTED = process.env.NEMOCLAW_E2E_USE_HOSTED_INFERENCE === "1";
 export const MODEL =
   process.env.NEMOCLAW_TURN_LATENCY_MODEL ??
diff --git a/test/e2e-scenario/live/hermes-discord.test.ts b/test/e2e-scenario/live/hermes-discord.test.ts
index 3926fab3254..d6ad565b889 100644
--- a/test/e2e-scenario/live/hermes-discord.test.ts
+++ b/test/e2e-scenario/live/hermes-discord.test.ts
@@ -47,11 +47,11 @@ function commandEnv(apiKey?: string, extra: NodeJS.ProcessEnv = {}): NodeJS.Proc
       NEMOCLAW_PROVIDER: process.env.NEMOCLAW_PROVIDER ?? "custom",
       NEMOCLAW_ENDPOINT_URL:
         process.env.NEMOCLAW_ENDPOINT_URL ?? "https://inference-api.nvidia.com/v1",
-      NEMOCLAW_MODEL: process.env.NEMOCLAW_MODEL ?? "nvidia/nemotron-3-ultra",
+      NEMOCLAW_MODEL: process.env.NEMOCLAW_MODEL ?? "nvidia/nvidia/nemotron-3-ultra",
       NEMOCLAW_COMPAT_MODEL:
         process.env.NEMOCLAW_COMPAT_MODEL ??
         process.env.NEMOCLAW_MODEL ??
-        "nvidia/nemotron-3-ultra",
+        "nvidia/nvidia/nemotron-3-ultra",
       NEMOCLAW_PREFERRED_API: process.env.NEMOCLAW_PREFERRED_API ?? "openai-completions",
       DISCORD_BOT_TOKEN: DISCORD_TOKEN,
       DISCORD_SERVER_IDS,
diff --git a/test/e2e-scenario/live/hermes-slack-e2e-helpers.ts b/test/e2e-scenario/live/hermes-slack-e2e-helpers.ts
index 8530f91519f..adf999f6d94 100644
--- a/test/e2e-scenario/live/hermes-slack-e2e-helpers.ts
+++ b/test/e2e-scenario/live/hermes-slack-e2e-helpers.ts
@@ -40,11 +40,11 @@ function hermesSlackEnv(apiKey?: string): NodeJS.ProcessEnv {
     apiKey,
     extra: {
       ...(apiKey ? { COMPATIBLE_API_KEY: apiKey } : {}),
-      NEMOCLAW_COMPAT_MODEL: process.env.NEMOCLAW_COMPAT_MODEL ?? "nvidia/nemotron-3-ultra",
+      NEMOCLAW_COMPAT_MODEL: process.env.NEMOCLAW_COMPAT_MODEL ?? "nvidia/nvidia/nemotron-3-ultra",
       NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1",
       NEMOCLAW_ENDPOINT_URL:
         process.env.NEMOCLAW_ENDPOINT_URL ?? "https://inference-api.nvidia.com/v1",
-      NEMOCLAW_MODEL: process.env.NEMOCLAW_MODEL ?? "nvidia/nemotron-3-ultra",
+      NEMOCLAW_MODEL: process.env.NEMOCLAW_MODEL ?? "nvidia/nvidia/nemotron-3-ultra",
       NEMOCLAW_POLICY_TIER: process.env.NEMOCLAW_POLICY_TIER ?? "open",
       NEMOCLAW_PREFERRED_API: process.env.NEMOCLAW_PREFERRED_API ?? "openai-completions",
       NEMOCLAW_PROVIDER: process.env.NEMOCLAW_PROVIDER ?? "custom",
diff --git a/test/e2e-scenario/live/rebuild-hermes.test.ts b/test/e2e-scenario/live/rebuild-hermes.test.ts
index 2127093379b..2df0fe7b052 100644
--- a/test/e2e-scenario/live/rebuild-hermes.test.ts
+++ b/test/e2e-scenario/live/rebuild-hermes.test.ts
@@ -54,7 +54,9 @@ const BACKUP_ROOT = path.join(os.homedir(), ".nemoclaw", "rebuild-backups");
 const HOSTED_ENDPOINT_URL =
   process.env.NEMOCLAW_ENDPOINT_URL ?? "https://inference-api.nvidia.com/v1";
 const HOSTED_MODEL =
-  process.env.NEMOCLAW_MODEL ?? process.env.NEMOCLAW_COMPAT_MODEL ?? "nvidia/nemotron-3-ultra";
+  process.env.NEMOCLAW_MODEL ??
+  process.env.NEMOCLAW_COMPAT_MODEL ??
+  "nvidia/nvidia/nemotron-3-ultra";
 const OLD_BASE_TAG = `nemoclaw-hermes-old-base:${SANDBOX_NAME.toLowerCase().replace(/[^a-z0-9_.-]+/g, "-")}`;
 const CURRENT_BASE_TAG = "ghcr.io/nvidia/nemoclaw/hermes-sandbox-base:latest";
 
diff --git a/test/e2e-scenario/live/upgrade-stale-sandbox-helpers.ts b/test/e2e-scenario/live/upgrade-stale-sandbox-helpers.ts
index 0b64cfd3a40..8b6fecc2ffb 100644
--- a/test/e2e-scenario/live/upgrade-stale-sandbox-helpers.ts
+++ b/test/e2e-scenario/live/upgrade-stale-sandbox-helpers.ts
@@ -122,7 +122,7 @@ export function writeStaleRegistryEntry(): void {
     (typeof session.model === "string" && session.model) ||
     process.env.NEMOCLAW_MODEL ||
     process.env.NEMOCLAW_COMPAT_MODEL ||
-    "nvidia/nemotron-3-ultra";
+    "nvidia/nvidia/nemotron-3-ultra";
   const registry = readJsonFile<{
     sandboxes?: Record<string, Record<string, unknown>>;
     defaultSandbox?: string;
diff --git a/test/e2e-script-workflow.test.ts b/test/e2e-script-workflow.test.ts
index 4b204933b18..f966b53f988 100644
--- a/test/e2e-script-workflow.test.ts
+++ b/test/e2e-script-workflow.test.ts
@@ -546,8 +546,8 @@ describe("E2E reusable workflow contract", () => {
     expect(runStep?.env?.NVIDIA_INFERENCE_API_KEY).toBe(GUARDED_HOSTED_INFERENCE_SECRET);
     expect(runStep?.env?.NEMOCLAW_PROVIDER).toBe("custom");
     expect(runStep?.env?.NEMOCLAW_ENDPOINT_URL).toBe("https://inference-api.nvidia.com/v1");
-    expect(runStep?.env?.NEMOCLAW_MODEL).toBe("nvidia/nemotron-3-ultra");
-    expect(runStep?.env?.NEMOCLAW_COMPAT_MODEL).toBe("nvidia/nemotron-3-ultra");
+    expect(runStep?.env?.NEMOCLAW_MODEL).toBe("nvidia/nvidia/nemotron-3-ultra");
+    expect(runStep?.env?.NEMOCLAW_COMPAT_MODEL).toBe("nvidia/nvidia/nemotron-3-ultra");
     expect(runStep?.env?.NEMOCLAW_PREFERRED_API).toBe("openai-completions");
     expect(runStep?.env?.COMPATIBLE_API_KEY).toBe(GUARDED_HOSTED_INFERENCE_SECRET);
     expect(runStep?.env?.GITHUB_TOKEN).toBeUndefined();
@@ -926,8 +926,8 @@ describe("E2E reusable workflow contract", () => {
     expect(exportStep?.run).toContain("NEMOCLAW_E2E_USE_HOSTED_INFERENCE=1");
     expect(exportStep?.run).toContain("NEMOCLAW_PROVIDER=custom");
     expect(exportStep?.run).toContain("NEMOCLAW_ENDPOINT_URL=https://inference-api.nvidia.com/v1");
-    expect(exportStep?.run).toContain("NEMOCLAW_MODEL=nvidia/nemotron-3-ultra");
-    expect(exportStep?.run).toContain("NEMOCLAW_COMPAT_MODEL=nvidia/nemotron-3-ultra");
+    expect(exportStep?.run).toContain("NEMOCLAW_MODEL=nvidia/nvidia/nemotron-3-ultra");
+    expect(exportStep?.run).toContain("NEMOCLAW_COMPAT_MODEL=nvidia/nvidia/nemotron-3-ultra");
     expect(exportStep?.run).toContain("NEMOCLAW_PREFERRED_API=openai-completions");
     expect(exportStep?.run).toContain("COMPATIBLE_API_KEY=%s");
 
@@ -950,7 +950,7 @@ describe("E2E reusable workflow contract", () => {
       expect(body, fixture).toContain("if env_provider == 'custom'");
       expect(body, fixture).toContain("'provider': provider");
       expect(body, fixture).toContain("'model': model");
-      expect(body, fixture).toContain("nvidia/nemotron-3-ultra");
+      expect(body, fixture).toContain("nvidia/nvidia/nemotron-3-ultra");
       expect(body, fixture).not.toContain("'provider': 'nvidia-prod'");
       expect(body, fixture).not.toContain("'model': 'nvidia/nemotron-3-super-120b-a12b'");
     }
@@ -1003,8 +1003,8 @@ describe("E2E reusable workflow contract", () => {
       }
       expect(step.env?.NEMOCLAW_PROVIDER, jobName).toBe("custom");
       expect(step.env?.NEMOCLAW_ENDPOINT_URL, jobName).toBe("https://inference-api.nvidia.com/v1");
-      expect(step.env?.NEMOCLAW_MODEL, jobName).toBe("nvidia/nemotron-3-ultra");
-      expect(step.env?.NEMOCLAW_COMPAT_MODEL, jobName).toBe("nvidia/nemotron-3-ultra");
+      expect(step.env?.NEMOCLAW_MODEL, jobName).toBe("nvidia/nvidia/nemotron-3-ultra");
+      expect(step.env?.NEMOCLAW_COMPAT_MODEL, jobName).toBe("nvidia/nvidia/nemotron-3-ultra");
       expect(step.env?.NEMOCLAW_PREFERRED_API, jobName).toBe("openai-completions");
       expect(step.env?.COMPATIBLE_API_KEY, jobName).toBe(GUARDED_HOSTED_INFERENCE_SECRET);
     }
diff --git a/test/e2e/lib/ci-compatible-inference.sh b/test/e2e/lib/ci-compatible-inference.sh
index 285da0389f7..12cdb94bcb1 100755
--- a/test/e2e/lib/ci-compatible-inference.sh
+++ b/test/e2e/lib/ci-compatible-inference.sh
@@ -7,7 +7,7 @@
 # at inference-api.nvidia.com. Keep this helper in test/e2e so the
 # product-facing provider/default endpoint remain unchanged.
 
-NEMOCLAW_E2E_COMPATIBLE_INFERENCE_MODEL_DEFAULT="nvidia/nemotron-3-ultra"
+NEMOCLAW_E2E_COMPATIBLE_INFERENCE_MODEL_DEFAULT="nvidia/nvidia/nemotron-3-ultra"
 NEMOCLAW_E2E_HOSTED_INFERENCE_PROVIDER_DEFAULT="compatible-endpoint"
 NEMOCLAW_E2E_NVIDIA_INFERENCE_MODEL_DEFAULT="nvidia/nemotron-3-super-120b-a12b"
 
diff --git a/test/e2e/test-rebuild-hermes.sh b/test/e2e/test-rebuild-hermes.sh
index 8183955dc05..f5d83266e91 100755
--- a/test/e2e/test-rebuild-hermes.sh
+++ b/test/e2e/test-rebuild-hermes.sh
@@ -261,7 +261,7 @@ model = (
     sess.get('model')
     or os.environ.get('NEMOCLAW_MODEL')
     or os.environ.get('NEMOCLAW_COMPAT_MODEL')
-    or 'nvidia/nemotron-3-ultra'
+    or 'nvidia/nvidia/nemotron-3-ultra'
 )
 credential_hash = hashlib.sha256('${DISCORD_FAKE_TOKEN}'.encode()).hexdigest()
 plan = {
@@ -397,7 +397,7 @@ fi
 
 # Inference works after rebuild (proves credential chain is intact)
 info "Verifying inference after rebuild..."
-POST_REBUILD_INFERENCE_MODEL="${NEMOCLAW_MODEL:-${NEMOCLAW_COMPAT_MODEL:-nvidia/nemotron-3-ultra}}"
+POST_REBUILD_INFERENCE_MODEL="${NEMOCLAW_MODEL:-${NEMOCLAW_COMPAT_MODEL:-nvidia/nvidia/nemotron-3-ultra}}"
 INFERENCE_RESPONSE=$(openshell sandbox exec --name "${SANDBOX_NAME}" -- \
   curl -s --max-time 60 https://inference.local/v1/chat/completions \
   -H 'Content-Type: application/json' \
diff --git a/test/e2e/test-rebuild-openclaw.sh b/test/e2e/test-rebuild-openclaw.sh
index 8aab09c93a0..7535602b3f9 100755
--- a/test/e2e/test-rebuild-openclaw.sh
+++ b/test/e2e/test-rebuild-openclaw.sh
@@ -230,7 +230,7 @@ model = (
     sess.get('model')
     or os.environ.get('NEMOCLAW_MODEL')
     or os.environ.get('NEMOCLAW_COMPAT_MODEL')
-    or 'nvidia/nemotron-3-ultra'
+    or 'nvidia/nvidia/nemotron-3-ultra'
 )
 reg = {'sandboxes': {'${SANDBOX_NAME}': {
     'name': '${SANDBOX_NAME}',
@@ -446,7 +446,7 @@ fi
 
 # Inference works after rebuild (proves credential chain is intact)
 info "Verifying inference after rebuild..."
-POST_REBUILD_INFERENCE_MODEL="${NEMOCLAW_MODEL:-${NEMOCLAW_COMPAT_MODEL:-nvidia/nemotron-3-ultra}}"
+POST_REBUILD_INFERENCE_MODEL="${NEMOCLAW_MODEL:-${NEMOCLAW_COMPAT_MODEL:-nvidia/nvidia/nemotron-3-ultra}}"
 INFERENCE_RESPONSE=$(openshell sandbox exec --name "${SANDBOX_NAME}" -- \
   curl -s --max-time 60 https://inference.local/v1/chat/completions \
   -H 'Content-Type: application/json' \
diff --git a/test/e2e/test-upgrade-stale-sandbox.sh b/test/e2e/test-upgrade-stale-sandbox.sh
index feb1bf6ab5a..e06c03d53b7 100755
--- a/test/e2e/test-upgrade-stale-sandbox.sh
+++ b/test/e2e/test-upgrade-stale-sandbox.sh
@@ -170,7 +170,7 @@ model = (
     sess.get('model')
     or os.environ.get('NEMOCLAW_MODEL')
     or os.environ.get('NEMOCLAW_COMPAT_MODEL')
-    or 'nvidia/nemotron-3-ultra'
+    or 'nvidia/nvidia/nemotron-3-ultra'
 )
 reg = {'sandboxes': {'${SANDBOX_NAME}': {
     'name': '${SANDBOX_NAME}',
diff --git a/test/issue-5667-hosted-inference-model-namespace.test.ts b/test/issue-5667-hosted-inference-model-namespace.test.ts
index 088d9cdac1b..a65cb025b70 100644
--- a/test/issue-5667-hosted-inference-model-namespace.test.ts
+++ b/test/issue-5667-hosted-inference-model-namespace.test.ts
@@ -1,12 +1,13 @@
 // SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 // SPDX-License-Identifier: Apache-2.0
 //
-// Regression test for issue #5667: onboarding a Deep Agents / OpenAI-compatible
-// sandbox without an explicit NEMOCLAW_MODEL recorded the default model id as
-// "nvidia/nvidia/nemotron-3-ultra" — a doubled "nvidia/" namespace prefix.
-// Standard NIM model ids carry exactly one namespace segment, so the default
-// fallback must be the canonical single-prefix id and the staged onboarding env
-// must never contain "nvidia/nvidia/".
+// Regression coverage for the hosted Inference Hub compatible-endpoint default.
+// The repo-secret endpoint at https://inference-api.nvidia.com/v1 is staged as
+// a custom OpenAI-compatible provider and expects provider/namespace/model IDs.
+// For NVIDIA-hosted models that means nvidia/nvidia/<model>, which is distinct
+// from the official NVIDIA provider catalog IDs used for build.nvidia.com.
+// NemoClaw must preserve the provider-accepted ID end-to-end instead of
+// normalizing away the leading provider segment.
 
 import assert from "node:assert/strict";
 import { spawnSync } from "node:child_process";
@@ -141,12 +142,12 @@ describe("issue #5667: hosted inference default model namespace", () => {
     Object.assign(process.env, envSnapshot);
   });
 
-  it("default hosted inference model has a single nvidia/ namespace segment", () => {
-    expect(providers.HOSTED_INFERENCE_MODEL).not.toContain("nvidia/nvidia/");
-    expect(providers.HOSTED_INFERENCE_MODEL).toBe("nvidia/nemotron-3-ultra");
+  it("default hosted inference model uses the provider/namespace/model convention", () => {
+    expect(providers.HOSTED_INFERENCE_MODEL).toBe("nvidia/nvidia/nemotron-3-ultra");
+    expect(providers.HOSTED_INFERENCE_MODEL).not.toContain("nvidia/nvidia/nvidia/");
   });
 
-  it("staging the hosted inference secret without NEMOCLAW_MODEL records a single-prefix model", () => {
+  it("staging the hosted inference secret without NEMOCLAW_MODEL records the provider-convention model", () => {
     // Reproduce the reported flow: an Inference Hub OpenAI-compatible key with no
     // explicit NEMOCLAW_MODEL, so onboarding falls back to the default model id.
     process.env.NVIDIA_INFERENCE_API_KEY = "sk-test-inference-hub-key";
@@ -155,12 +156,12 @@ describe("issue #5667: hosted inference default model namespace", () => {
     const staged = providers.stageHostedInferenceSourceSecretEnv();
 
     expect(staged).toBe(true);
-    expect(process.env.NEMOCLAW_MODEL).not.toContain("nvidia/nvidia/");
-    expect(process.env.NEMOCLAW_MODEL).toBe("nvidia/nemotron-3-ultra");
-    expect(process.env.NEMOCLAW_COMPAT_MODEL).toBe("nvidia/nemotron-3-ultra");
+    expect(process.env.NEMOCLAW_MODEL).toBe("nvidia/nvidia/nemotron-3-ultra");
+    expect(process.env.NEMOCLAW_MODEL).not.toContain("nvidia/nvidia/nvidia/");
+    expect(process.env.NEMOCLAW_COMPAT_MODEL).toBe("nvidia/nvidia/nemotron-3-ultra");
   });
 
-  it("stages the Deep Agents NEMOCLAW_PROVIDER_KEY path with a single-prefix model", () => {
+  it("stages the Deep Agents NEMOCLAW_PROVIDER_KEY path with the provider-convention model", () => {
     // Reproduce the issue command: a Deep Agents compatible endpoint key is
     // supplied via the generic provider-key hint, with no explicit model.
     process.env.NEMOCLAW_AGENT = "langchain-deepagents-code";
@@ -171,12 +172,12 @@ describe("issue #5667: hosted inference default model namespace", () => {
     expect(staged).toBe(true);
     expect(process.env.NEMOCLAW_PROVIDER).toBe("custom");
     expect(process.env.COMPATIBLE_API_KEY).toBe("sk-test-inference-hub-key");
-    expect(process.env.NEMOCLAW_MODEL).not.toContain("nvidia/nvidia/");
-    expect(process.env.NEMOCLAW_MODEL).toBe("nvidia/nemotron-3-ultra");
-    expect(process.env.NEMOCLAW_COMPAT_MODEL).toBe("nvidia/nemotron-3-ultra");
+    expect(process.env.NEMOCLAW_MODEL).toBe("nvidia/nvidia/nemotron-3-ultra");
+    expect(process.env.NEMOCLAW_MODEL).not.toContain("nvidia/nvidia/nvidia/");
+    expect(process.env.NEMOCLAW_COMPAT_MODEL).toBe("nvidia/nvidia/nemotron-3-ultra");
   });
 
-  it("drives setupNim and downstream Deep Agents surfaces with a single-prefix model", async () => {
+  it("drives setupNim and downstream Deep Agents surfaces with the provider-convention model", async () => {
     const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-issue-5667-"));
     const fakeBin = path.join(tmpDir, "bin");
     const home = path.join(tmpDir, "home");
@@ -253,16 +254,16 @@ const { setupNim } = require(${onboardPath});
 
       expect(payload.result.provider).toBe("compatible-endpoint");
       expect(payload.result.credentialEnv).toBe("COMPATIBLE_API_KEY");
-      expect(payload.result.model).toBe("nvidia/nemotron-3-ultra");
+      expect(payload.result.model).toBe("nvidia/nvidia/nemotron-3-ultra");
       expect(payload.result.preferredInferenceApi).toBe("openai-completions");
       expect(payload.env).toMatchObject({
         provider: "custom",
-        model: "nvidia/nemotron-3-ultra",
-        compatModel: "nvidia/nemotron-3-ultra",
+        model: "nvidia/nvidia/nemotron-3-ultra",
+        compatModel: "nvidia/nvidia/nemotron-3-ultra",
         compatibleKey: "sk-test-inference-hub-key",
         preferredApi: "openai-completions",
       });
-      expect(output).not.toContain("nvidia/nvidia/");
+      expect(output).not.toContain("nvidia/nvidia/nvidia/");
 
       const statusSnapshot = await collectSandboxStatusSnapshot("dcode-test", {
         deps: {
@@ -278,15 +279,20 @@ const { setupNim } = require(${onboardPath});
       });
       const statusModelLine = `    Model:    ${statusSnapshot.currentModel}`;
       expect(statusSnapshot.currentProvider).toBe("compatible-endpoint");
-      expect(statusSnapshot.currentModel).toBe("nvidia/nemotron-3-ultra");
-      expect(statusModelLine).toBe("    Model:    nvidia/nemotron-3-ultra");
-      expect(statusModelLine).not.toContain("nvidia/nvidia/");
+      expect(statusSnapshot.currentModel).toBe("nvidia/nvidia/nemotron-3-ultra");
+      expect(statusModelLine).toBe("    Model:    nvidia/nvidia/nemotron-3-ultra");
+      expect(statusModelLine).not.toContain("nvidia/nvidia/nvidia/");
 
       const dockerfilePath = path.join(tmpDir, "Dockerfile");
       fs.writeFileSync(dockerfilePath, "FROM scratch\nARG NEMOCLAW_MODEL=old\n");
-      patchStagedDockerfile(dockerfilePath, payload.result.model, null, "issue-5667-single-prefix");
+      patchStagedDockerfile(
+        dockerfilePath,
+        payload.result.model,
+        null,
+        "issue-5667-provider-convention",
+      );
       expect(fs.readFileSync(dockerfilePath, "utf8")).toContain(
-        "ARG NEMOCLAW_MODEL=nvidia/nemotron-3-ultra",
+        "ARG NEMOCLAW_MODEL=nvidia/nvidia/nemotron-3-ultra",
       );
 
       const configResult = spawnSync(
@@ -312,8 +318,8 @@ const { setupNim } = require(${onboardPath});
       );
       assert.equal(configResult.status, 0, `${configResult.stdout}\n${configResult.stderr}`);
       const config = fs.readFileSync(path.join(home, ".deepagents", "config.toml"), "utf8");
-      expect(config).toContain('default = "openai:nvidia/nemotron-3-ultra"');
-      expect(config).not.toContain("nvidia/nvidia/");
+      expect(config).toContain('default = "openai:nvidia/nvidia/nemotron-3-ultra"');
+      expect(config).not.toContain("nvidia/nvidia/nvidia/");
 
       const dcodeWrapperPath = writeDcodeWrapperFixture(tmpDir, home);
       const fakePythonPath = writeFakeDeepAgentsCodeModule(tmpDir);
@@ -329,10 +335,10 @@ const { setupNim } = require(${onboardPath});
       const dcodeOutput = `${dcodeResult.stdout}\n${dcodeResult.stderr}`;
       assert.equal(dcodeResult.status, 0, dcodeOutput);
       expect(dcodeOutput).toContain(
-        "App: v0.1.12 | Agent: agent (default) | Model: nvidia/nemotron-3-ultra",
+        "App: v0.1.12 | Agent: agent (default) | Model: nvidia/nvidia/nemotron-3-ultra",
       );
       expect(dcodeOutput).toContain("ARGS:--sandbox none --no-mcp -n ping");
-      expect(dcodeOutput).not.toContain("nvidia/nvidia/");
+      expect(dcodeOutput).not.toContain("nvidia/nvidia/nvidia/");
     } finally {
       fs.rmSync(tmpDir, { recursive: true, force: true });
     }
diff --git a/tools/e2e-scenarios/workflow-boundary.mts b/tools/e2e-scenarios/workflow-boundary.mts
index 71d1ac85a28..6811e3bb65a 100644
--- a/tools/e2e-scenarios/workflow-boundary.mts
+++ b/tools/e2e-scenarios/workflow-boundary.mts
@@ -1771,10 +1771,10 @@ function validateRebuildHermesVitestJob(
   if (jobEnv.NEMOCLAW_ENDPOINT_URL !== "https://inference-api.nvidia.com/v1") {
     errors.push(`${jobName} job must target hosted CI inference endpoint`);
   }
-  if (jobEnv.NEMOCLAW_MODEL !== "nvidia/nemotron-3-ultra") {
+  if (jobEnv.NEMOCLAW_MODEL !== "nvidia/nvidia/nemotron-3-ultra") {
     errors.push(`${jobName} job must pin the CI-safe Hermes rebuild model`);
   }
-  if (jobEnv.NEMOCLAW_COMPAT_MODEL !== "nvidia/nemotron-3-ultra") {
+  if (jobEnv.NEMOCLAW_COMPAT_MODEL !== "nvidia/nvidia/nemotron-3-ultra") {
     errors.push(`${jobName} job must pin the CI-safe compatible model`);
   }
   if (jobEnv.OPENSHELL_GATEWAY !== "nemoclaw") {