Add Claude Opus 5.5 to the model registry (#170)

Anthropic released Opus 5.5 (2026-09-22), which beats Opus 5 on agentic
coding/knowledge-work benchmarks at 20% lower cost ($4/$20 vs $5/$25 per
MTok). Registered under the Anthropic section with supports_temperature=False
(adaptive thinking is always on, like every Opus 4.7+ model).

Companion catalog PR: OpenGradient/chat-api (adds the model to the served
catalog and hides the now-superseded Opus 5).


Claude-Session: https://claude.ai/code/session_015GBZNoRJrk2hPkEdYhDVHB

Co-authored-by: Claude <noreply@anthropic.com>

Author: adambalogh

CLAUDE.md

@@ -127,7 +127,7 @@ Server configuration:
 `model_registry.py` is the source of truth; this list mirrors it. Model name
 prefixes determine routing:
 - **OpenAI**: gpt-6-astra, gpt-4.1, gpt-4.1-mini, gpt-5, gpt-5-mini, gpt-5.2, gpt-5.4, gpt-5.4-mini, gpt-5.4-nano, gpt-5.5, gpt-5.6-sol/terra/luna, o3; image generation: gpt-image-2.5-flare, gpt-image-2.5-sunburst, gpt-image-2
-- **Anthropic**: claude-sonnet-4-5/4-6, claude-sonnet-5, claude-haiku-4-5, claude-opus-4-5/4-6/4-7/4-8, claude-opus-5, claude-fable-5, claude-fable-5-1
+- **Anthropic**: claude-sonnet-4-5/4-6, claude-sonnet-5, claude-haiku-4-5, claude-opus-4-5/4-6/4-7/4-8, claude-opus-5, claude-opus-5-5, claude-fable-5, claude-fable-5-1
 - **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-4.7, 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

README.md

@@ -34,7 +34,7 @@ The gateway solves this by running inside a hardware-isolated Nitro Enclave wher
 | Provider | Models |
 |----------|--------|
 | OpenAI | gpt-6-astra, gpt-4.1, gpt-4.1-mini, gpt-5, gpt-5-mini, gpt-5.2, gpt-5.4, gpt-5.5, gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna, o3 |
-| Anthropic | claude-fable-5-1, claude-fable-5, claude-opus-5, claude-opus-4-8, claude-opus-4-7, claude-opus-4-6, claude-opus-4-5, claude-sonnet-5, claude-sonnet-4-6, claude-sonnet-4-5, claude-haiku-4-5 |
+| Anthropic | claude-fable-5-1, claude-fable-5, claude-opus-5-5, claude-opus-5, claude-opus-4-8, claude-opus-4-7, claude-opus-4-6, claude-opus-4-5, claude-sonnet-5, claude-sonnet-4-6, claude-sonnet-4-5, claude-haiku-4-5 |
 | 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-4.7, 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 |

tee_gateway/model_registry.py

@@ -540,6 +540,18 @@ class SupportedModel(Enum):
         output_price_usd=Decimal("0.000025"),
         supports_temperature=False,
     )
+    # Claude Opus 5.5 — successor to Opus 5, GA 2026-09-22. Leads on agentic
+    # coding/knowledge-work benchmarks while costing 20% less than Opus 5
+    # ($4/$20 vs $5/$25 per MTok). Adaptive-thinking-only (always on, can't be
+    # disabled); like Opus 4.7+ it rejects the `temperature` field (HTTP 400),
+    # so supports_temperature=False.
+    CLAUDE_OPUS_5_5 = ModelConfig(
+        provider="anthropic",
+        api_name="claude-opus-5-5",
+        input_price_usd=Decimal("0.000004"),
+        output_price_usd=Decimal("0.00002"),
+        supports_temperature=False,
+    )
     # Claude Opus 5 — the current Opus, a drop-in upgrade at Opus 4.8's pricing
     # ($5/$25 per MTok). Adaptive-thinking-only; like Opus 4.7+ it rejects the
     # `temperature` field (HTTP 400), so supports_temperature=False.
@@ -1067,6 +1079,8 @@ class SupportedModel(Enum):
     "claude-opus-4-7": SupportedModel.CLAUDE_OPUS_4_7,
     "claude-opus-4-8": SupportedModel.CLAUDE_OPUS_4_8,
     "claude-opus-5": SupportedModel.CLAUDE_OPUS_5,
+    "claude-opus-5-5": SupportedModel.CLAUDE_OPUS_5_5,
+    "claude-opus-5.5": SupportedModel.CLAUDE_OPUS_5_5,
     "claude-fable-5": SupportedModel.CLAUDE_FABLE_5,
     "claude-fable-5-1": SupportedModel.CLAUDE_FABLE_5_1,
     "claude-fable-5.1": SupportedModel.CLAUDE_FABLE_5_1,

tests/test_pricing.py

@@ -136,6 +136,19 @@ def test_claude_opus_5_resolves(self):
         # Adaptive-thinking-only; rejects the `temperature` field (HTTP 400)
         self.assertFalse(cfg.supports_temperature)
 
+    def test_claude_opus_5_5_resolves(self):
+        cfg = get_model_config("claude-opus-5-5")
+        self.assertEqual(cfg.provider, "anthropic")
+        self.assertEqual(cfg.api_name, "claude-opus-5-5")
+        self.assertEqual(cfg.input_price_usd, Decimal("0.000004"))
+        self.assertEqual(cfg.output_price_usd, Decimal("0.00002"))
+        # Adaptive-thinking-only; rejects the `temperature` field (HTTP 400)
+        self.assertFalse(cfg.supports_temperature)
+
+    def test_claude_opus_5_5_dotted_alias_resolves(self):
+        cfg = get_model_config("claude-opus-5.5")
+        self.assertEqual(cfg, get_model_config("claude-opus-5-5"))
+
     def test_claude_fable_5_resolves(self):
         cfg = get_model_config("claude-fable-5")
         self.assertEqual(cfg.provider, "anthropic")
@@ -820,6 +833,13 @@ def test_claude_opus_5_cost(self):
         # Same price tier as opus-4-5/4-6/4-7/4-8: 1000*0.000005 + 500*0.000025 = 0.0175 USD
         self.assertEqual(cost, 17_500_000_000_000_000)
 
+    def test_claude_opus_5_5_cost(self):
+        cost = self._calc("claude-opus-5-5", 1000, 500)
+        expected = _expected_cost_opg("claude-opus-5-5", 1000, 500)
+        self.assertEqual(cost, expected)
+        # 1000*0.000004 + 500*0.00002 = 0.004 + 0.01 = 0.014 USD = 1.4e16 wei
+        self.assertEqual(cost, 14_000_000_000_000_000)
+
     # ── Anthropic Fable ─────────────────────────────────────────────────────
 
     def test_claude_fable_5_1_cost(self):