Remove the delisted hermes-4-70b; charge OpenRouter's real rate for hy3

CLAUDE.md

@@ -131,7 +131,7 @@ prefixes determine routing:
 - **Google**: gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-3.5-flash, gemini-3.5-flash-lite, gemini-2.5-flash, gemini-2.5-flash-lite, gemini-2.5-pro, gemini-3-flash-preview, gemini-3.1-pro-preview; image generation: gemini-2.5-flash-image, gemini-3.1-flash-image
 - **xAI**: grok-4.6, grok-4.5, grok-4.3, grok-4.20-reasoning, grok-4.20-non-reasoning; image generation: grok-2-image, grok-imagine-image-2.0
 - **ByteDance** (BytePlus ModelArk, OpenAI-compatible, ap-southeast): seed-1.6, seed-1.8, seed-2.0-lite, deepseek-v4-flash, deepseek-v4-pro, glm-5.2 (Z.ai's model served via a ModelArk deployment endpoint); image generation: seedream-4.0, seedream-5.0-lite, seedance-4.5, seedance-5.0
-- **OpenRouter** (OpenAI-compatible): hy4-preview, hermes-4-405b, hermes-4-70b, hy3
+- **OpenRouter** (OpenAI-compatible): hy4-preview, hermes-4-405b, hy3
 - **Z.ai** (Model API, OpenAI-compatible): image generation: glm-image (glm-5.2 chat is routed through BytePlus ModelArk, see ByteDance above)
 
 Models kept registered but no longer offered to new clients — each still
@@ -149,6 +149,12 @@ provider actually bills, not at its pre-retirement rate:
   repointed or retired with its snapshot; if repointed, they become mispriced
   the way the xAI slugs were.
 
+A model the provider does not serve at all is removed outright rather than
+repriced — no request to it can succeed. `hermes-4-70b` went this way when
+OpenRouter delisted it. For OpenRouter models, `openrouter.ai/api/v1/models`
+is authoritative for availability as well as price; a pricing page keeps
+quoting a rate after the endpoints are gone.
+
 Image generation via OpenAI (gpt-image-2.5-flare, gpt-image-2.5-sunburst, gpt-image-2), xAI (grok-2-image, grok-imagine-image-2.0), ByteDance
 (seedream-4.0, seedream-5.0-lite, seedance-4.5, seedance-5.0), and Z.ai (glm-image) is served
 through a provider `/images/generations` endpoint rather than the chat path (see

README.md

@@ -38,7 +38,7 @@ The gateway solves this by running inside a hardware-isolated Nitro Enclave wher
 | Google | gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-3.5-flash, gemini-3.5-flash-lite, gemini-2.5-flash, gemini-2.5-flash-lite, gemini-2.5-pro, gemini-3-flash-preview, gemini-3.1-pro-preview |
 | xAI | grok-4.6, grok-4.5, grok-4.3, grok-4.20-reasoning, grok-4.20-non-reasoning |
 | ByteDance | seed-1.6, seed-1.8, seed-2.0-lite, dola-seed-2.0-mini, deepseek-v4-flash, deepseek-v4-pro, glm-5.2 |
-| OpenRouter | hy4-preview, hermes-4-405b, hermes-4-70b, hy3 |
+| OpenRouter | hy4-preview, hermes-4-405b, hy3 |
 
 Image generation: gpt-image-2, gpt-image-2.5-flare, gpt-image-2.5-sunburst,
 gemini-2.5-flash-image, gemini-3.1-flash-image, grok-2-image,

tee_gateway/model_registry.py

@@ -939,25 +939,24 @@ class SupportedModel(Enum):
         input_price_usd=Decimal("0.000000834"),
         output_price_usd=Decimal("0.000002501"),
     )
-    # Nous no longer serves Hermes 4 through Nous Portal, so both models route
-    # through their canonical OpenRouter slugs and use OpenRouter list pricing.
+    # Nous no longer serves Hermes 4 through Nous Portal, so it routes through
+    # its canonical OpenRouter slug and uses OpenRouter list pricing. Only the
+    # 405B is served — OpenRouter has delisted hermes-4-70b. A model's pricing
+    # page outlives its endpoints, so check /api/v1/models for availability
+    # rather than trusting a quoted rate.
     HERMES_4_405B = ModelConfig(
         provider="openrouter",
         api_name="nousresearch/hermes-4-405b",
         input_price_usd=Decimal("0.000001"),
         output_price_usd=Decimal("0.000003"),
     )
-    HERMES_4_70B = ModelConfig(
-        provider="openrouter",
-        api_name="nousresearch/hermes-4-70b",
-        input_price_usd=Decimal("0.00000013"),
-        output_price_usd=Decimal("0.0000004"),
-    )
+    # Rate from openrouter.ai/api/v1/models, which is authoritative for both
+    # price and availability.
     HY3 = ModelConfig(
         provider="openrouter",
         api_name="tencent/hy3",
-        input_price_usd=Decimal("0.0000000825"),
-        output_price_usd=Decimal("0.00000033"),
+        input_price_usd=Decimal("0.000000132"),
+        output_price_usd=Decimal("0.000000528"),
     )
 
     # ── Z.ai (Model API, OpenAI-compatible) ─────────────────────────────
@@ -1118,8 +1117,6 @@ class SupportedModel(Enum):
     # Nous Research models routed through OpenRouter
     "hermes-4-405b": SupportedModel.HERMES_4_405B,
     "nousresearch/hermes-4-405b": SupportedModel.HERMES_4_405B,
-    "hermes-4-70b": SupportedModel.HERMES_4_70B,
-    "nousresearch/hermes-4-70b": SupportedModel.HERMES_4_70B,
     "hy3": SupportedModel.HY3,
     "tencent/hy3": SupportedModel.HY3,
     "tencent/hy3:floor": SupportedModel.HY3,

tee_gateway/test/test_provider_usage_integration.py

@@ -52,9 +52,9 @@ class _ProviderCase:
     _ProviderCase("OpenAI", "gpt-4.1-nano", "OPENAI_API_KEY"),
     _ProviderCase("Anthropic", "claude-haiku-4-5", "ANTHROPIC_API_KEY"),
     _ProviderCase("Google", "gemini-3.5-flash-lite", "GOOGLE_API_KEY"),
-    _ProviderCase("xAI", "grok-4-fast", "XAI_API_KEY"),
+    _ProviderCase("xAI", "grok-4.3", "XAI_API_KEY"),
     _ProviderCase("ByteDance", "deepseek-v4-flash", "ARK_API_KEY"),
-    _ProviderCase("OpenRouter", "hermes-4-70b", "OPENROUTER_API_KEY"),
+    _ProviderCase("OpenRouter", "hy3", "OPENROUTER_API_KEY"),
     _ProviderCase("Z.ai", "glm-5.2", "ZAI_API_KEY"),
 )
 

tests/test_pricing.py

@@ -578,29 +578,30 @@ def test_hermes_4_405b_resolves(self):
         self.assertEqual(cfg.input_price_usd, Decimal("0.000001"))
         self.assertEqual(cfg.output_price_usd, Decimal("0.000003"))
 
-    def test_hermes_4_70b_resolves(self):
-        cfg = get_model_config("hermes-4-70b")
-        self.assertEqual(cfg.provider, "openrouter")
-        self.assertEqual(cfg.api_name, "nousresearch/hermes-4-70b")
-        self.assertEqual(cfg.input_price_usd, Decimal("0.00000013"))
-        self.assertEqual(cfg.output_price_usd, Decimal("0.0000004"))
+    def test_hermes_4_70b_is_gone(self):
+        """OpenRouter delisted nousresearch/hermes-4-70b.
+
+        It is absent from openrouter.ai/api/v1/models, so every request
+        returned "404 No endpoints found". Rejecting it here beats forwarding
+        a request that cannot succeed.
+        """
+        for name in ("hermes-4-70b", "nousresearch/hermes-4-70b"):
+            with self.subTest(model=name):
+                with self.assertRaises(ValueError):
+                    get_model_config(name)
 
     def test_openrouter_canonical_hermes_aliases_resolve(self):
         self.assertEqual(
             get_model_config("nousresearch/hermes-4-405b"),
             get_model_config("hermes-4-405b"),
         )
-        self.assertEqual(
-            get_model_config("nousresearch/hermes-4-70b"),
-            get_model_config("hermes-4-70b"),
-        )
 
     def test_hy3_resolves(self):
         cfg = get_model_config("hy3")
         self.assertEqual(cfg.provider, "openrouter")
         self.assertEqual(cfg.api_name, "tencent/hy3")
-        self.assertEqual(cfg.input_price_usd, Decimal("0.0000000825"))
-        self.assertEqual(cfg.output_price_usd, Decimal("0.00000033"))
+        self.assertEqual(cfg.input_price_usd, Decimal("0.000000132"))
+        self.assertEqual(cfg.output_price_usd, Decimal("0.000000528"))
         self.assertEqual(cfg, get_model_config("tencent/hy3"))
         self.assertEqual(cfg, get_model_config("tencent/hy3:floor"))
 
@@ -986,9 +987,10 @@ def test_deepseek_v4_pro_cost(self):
     # ── OpenRouter ─────────────────────────────────────────────────────────
 
     def test_hy3_cost(self):
+        # 1000*0.000000132 + 500*0.000000528 = 0.000132 + 0.000264 = 0.000396 USD
         self.assertEqual(
             self._calc("hy3", 1000, 500),
-            247_500_000_000_000,
+            396_000_000_000_000,
         )
 
     def test_hy4_preview_cost(self):