diff --git a/services/hermes/scripts/suite_backends.py b/services/hermes/scripts/suite_backends.py
index 5d91dac5..e2f7a79c 100644
--- a/services/hermes/scripts/suite_backends.py
+++ b/services/hermes/scripts/suite_backends.py
@@ -4,6 +4,7 @@ from __future__ import annotations
 import json
 import os
 from pathlib import Path
+import re
 import signal
 import subprocess
 import tempfile
@@ -129,6 +130,7 @@ def stop(process):
 def parse_claude(raw, expected_model):
     """Normalize only a completed, uncompacted CLI result with a pinned model."""
     final, initialized = None, False
+    aliases = {expected_model, expected_model + "[1m]"}
     for line in raw.splitlines():
         try:
             event = json.loads(line)
@@ -140,7 +142,7 @@ def parse_claude(raw, expected_model):
             raise Problem("compaction_detected", 502)
         if event.get("type") == "system" and event.get("subtype") == "init":
             initialized = True
-            if (event.get("model") != expected_model or event.get("mcp_servers") or
+            if (event.get("model") not in aliases or event.get("mcp_servers") or
                     event.get("plugins") or set(event.get("tools", [])) - {"StructuredOutput"}):
                 raise Problem("worker_isolation_failed", 502)
         if event.get("type") == "result":
@@ -156,11 +158,15 @@ def parse_claude(raw, expected_model):
     if final.get("stop_reason") in {"max_tokens", "model_context_window_exceeded"}:
         raise Problem("incomplete_generation", 502)
     models = final.get("modelUsage", {})
-    if set(models) != {expected_model}:
-        raise Problem("model_changed", 502)
-    limits = models[expected_model]
-    if limits.get("contextWindow") != 1000000 or limits.get("maxOutputTokens") != 64000:
-        raise Problem("backend_capabilities_changed", 502)
+    if not models or set(models) - aliases:
+        observed = [name if re.fullmatch(r"claude-[a-z0-9.-]+(?:\[1m\])?", name)
+                    else "unrecognized" for name in models]
+        raise Problem("model_changed", 502, observed_models=observed)
+    for limits in models.values():
+        if (limits.get("contextWindow") != 1000000 or limits.get("maxOutputTokens") != 64000
+                or limits.get("canonicalModel") != expected_model
+                or limits.get("provider") != "firstParty"):
+            raise Problem("backend_capabilities_changed", 502)
     usage = final.get("usage", {})
     if any(usage.get("server_tool_use", {}).values()):
         raise Problem("worker_isolation_failed", 502)
diff --git a/services/hermes/suite-planner-deployment.yaml b/services/hermes/suite-planner-deployment.yaml
index 70513be2..d4cc7a6d 100644
--- a/services/hermes/suite-planner-deployment.yaml
+++ b/services/hermes/suite-planner-deployment.yaml
@@ -36,7 +36,7 @@ spec:
         app: hermes-suite-planner
       annotations:
         fluentbit.io/exclude: "true"
-        ai.bstein.dev/config-rev: suite-v2-20260929
+        ai.bstein.dev/config-rev: suite-v3-20260929
         vault.hashicorp.com/agent-inject: "true"
         vault.hashicorp.com/agent-pre-populate-only: "true"
         vault.hashicorp.com/agent-init-first: "true"