fix(cache): route Nous Portal Qwen through Portal-Claude cache pathway

run_agent.py

@@ -3465,6 +3465,15 @@ def _anthropic_prompt_cache_policy(
             return True, True
         if (is_openrouter or is_nous_portal) and is_claude:
             return True, False
+        # Nous Portal Qwen (e.g. qwen3.6-plus) takes the same envelope-layout
+        # cache_control path as Portal Claude. Portal proxies to OpenRouter
+        # and the upstream Qwen route accepts cache_control markers; without
+        # this branch the alibaba-family check below only matches
+        # provider=opencode/alibaba and Portal traffic falls through to
+        # (False, False), serving 0% cache hits and re-billing the full
+        # prompt on every turn.
+        if is_nous_portal and "qwen" in model_lower:
+            return True, False
         if is_anthropic_wire and is_claude:
             # Third-party Anthropic-compatible gateway.
             return True, True
@@ -3540,7 +3549,19 @@ def _supports_long_lived_anthropic_cache(
         eff_api_mode = api_mode if api_mode is not None else (self.api_mode or "")
         eff_model = (model if model is not None else self.model) or ""
 
-        if "claude" not in eff_model.lower():
+        model_lower = eff_model.lower()
+        is_claude = "claude" in model_lower
+        is_nous_portal = "nousresearch" in eff_base_url.lower()
+
+        # Nous Portal: Claude AND Qwen both get long-lived caching.
+        # Portal proxies to OpenRouter with identical cache_control
+        # semantics; any model on Portal that accepts envelope-layout
+        # markers via _anthropic_prompt_cache_policy also benefits from
+        # the documented 1h cross-session TTL.
+        if is_nous_portal and (is_claude or "qwen" in model_lower):
+            return True
+
+        if not is_claude:
             return False
 
         # Native Anthropic + Anthropic OAuth subscription
@@ -3554,7 +3575,7 @@ def _supports_long_lived_anthropic_cache(
 
         # Nous Portal — front-ends OpenRouter behind the scenes; identical
         # wire format and cache_control semantics.
-        if "nousresearch" in eff_base_url.lower():
+        if is_nous_portal:
             return True
 
         return False

tests/run_agent/test_anthropic_prompt_cache_policy.py

@@ -257,6 +257,40 @@ def test_qwen_on_openrouter_not_affected(self):
         )
         assert agent._anthropic_prompt_cache_policy() == (False, False)
 
+    def test_qwen_on_nous_portal_caches_with_envelope_layout(self):
+        # Nous Portal Qwen takes the same envelope-layout cache_control
+        # path as Portal Claude. Without this, Portal-routed qwen3.6-plus
+        # falls through to the alibaba-family check (which only matches
+        # provider=opencode/alibaba) and serves 0% cache hits.
+        agent = _make_agent(
+            provider="nous",
+            base_url="https://inference-api.nousresearch.com/v1",
+            api_mode="chat_completions",
+            model="qwen3.6-plus",
+        )
+        assert agent._anthropic_prompt_cache_policy() == (True, False)
+
+    def test_qwen_vendored_slug_on_nous_portal_caches(self):
+        # Same path but with the vendored slug form Portal sometimes uses.
+        agent = _make_agent(
+            provider="nous",
+            base_url="https://inference-api.nousresearch.com/v1",
+            api_mode="chat_completions",
+            model="qwen/qwen3.6-plus",
+        )
+        assert agent._anthropic_prompt_cache_policy() == (True, False)
+
+    def test_non_qwen_non_claude_on_nous_portal_does_not_cache(self):
+        # Portal scope is narrow: Claude OR Qwen only. Other models
+        # routed through Portal keep their existing fall-through behavior.
+        agent = _make_agent(
+            provider="nous",
+            base_url="https://inference-api.nousresearch.com/v1",
+            api_mode="chat_completions",
+            model="openai/gpt-5.4",
+        )
+        assert agent._anthropic_prompt_cache_policy() == (False, False)
+
 
 class TestExplicitOverrides:
     """Policy accepts keyword overrides for switch_model / fallback activation."""
@@ -338,6 +372,37 @@ def test_nous_portal_claude_supported(self):
         )
         assert agent._supports_long_lived_anthropic_cache() is True
 
+    def test_nous_portal_qwen_supported(self):
+        # Portal Qwen rides the same OpenRouter-equivalent transport as
+        # Portal Claude; long-lived (1h cross-session) cache_control
+        # markers apply identically.
+        agent = _make_agent(
+            provider="nous",
+            base_url="https://inference-api.nousresearch.com/v1",
+            api_mode="chat_completions",
+            model="qwen3.6-plus",
+        )
+        assert agent._supports_long_lived_anthropic_cache() is True
+
+    def test_nous_portal_qwen_vendored_slug_supported(self):
+        agent = _make_agent(
+            provider="nous",
+            base_url="https://inference-api.nousresearch.com/v1",
+            api_mode="chat_completions",
+            model="qwen/qwen3.6-plus",
+        )
+        assert agent._supports_long_lived_anthropic_cache() is True
+
+    def test_nous_portal_non_claude_non_qwen_rejected(self):
+        # Portal long-lived cache scope mirrors policy: Claude or Qwen only.
+        agent = _make_agent(
+            provider="nous",
+            base_url="https://inference-api.nousresearch.com/v1",
+            api_mode="chat_completions",
+            model="openai/gpt-5.4",
+        )
+        assert agent._supports_long_lived_anthropic_cache() is False
+
     def test_openrouter_non_claude_rejected(self):
         agent = _make_agent(
             provider="openrouter",