feat: upgrade default agent + reviewer model to Opus 4.8 (#1350)

* feat: upgrade default agent + reviewer model to Opus 4.8

Replace Opus 4.7 with Opus 4.8 (claude-opus-4-8) as the supported
Anthropic model surfaced in the profile editor and used by the main
agent and reviewer graphs. Effort levels (low/medium/high/xhigh/max)
and the high default are unchanged, matching the official Opus 4.8
docs. Updates eval config comment and tests accordingly.

* fix: provider-aware fallback for stale stored model ids

Dropping claude-opus-4-7 from the supported set meant persisted
profile/team-settings still holding it failed the SUPPORTED_MODEL_IDS
check and fell through to default_model_pair() — a cross-provider jump
to the OpenAI global default.

Add provider_fallback_pair: when a stored id is no longer supported but
its provider still has a supported model, resolve to that provider's
newest supported model (anthropic:claude-opus-4-7 -> 4.8), preserving
effort when valid. Resolution order is now: valid stored pair ->
same-provider fallback -> global default_model_pair(). Profile overrides
keep deferring to the team default when no model is set or the provider
is unknown.

agent/dashboard/agent_overrides.py

@@ -9,7 +9,7 @@ import httpx
 from langgraph_sdk import get_client
 
 from ..utils.github_user_email_map import GITHUB_USER_EMAIL_MAP
-from .options import SUPPORTED_MODEL_IDS, model_supports_effort
+from .options import SUPPORTED_MODEL_IDS, model_supports_effort, provider_fallback_pair
 from .profiles import PROFILES_NAMESPACE
 
 logger = logging.getLogger(__name__)
@@ -97,6 +97,14 @@ def _normalize_profile_model_pair(
         and model_supports_effort(model_id, effort)
     ):
         return model_id, effort
+    # A stored selection whose exact id dropped out of the supported set (e.g. an
+    # Opus minor-version bump) stays on its provider rather than being discarded
+    # and silently deferring to the team default. An absent/unknown-provider
+    # selection still returns (None, None) so the team default applies.
+    if isinstance(model_id, str):
+        provider_pair = provider_fallback_pair(model_id, effort)
+        if provider_pair is not None:
+            return provider_pair
     return None, None
 
 

agent/dashboard/options.py

@@ -14,8 +14,8 @@ class ModelOption(TypedDict):
 
 SUPPORTED_MODELS: list[ModelOption] = [
     {
-        "id": "anthropic:claude-opus-4-7",
-        "label": "Opus 4.7",
+        "id": "anthropic:claude-opus-4-8",
+        "label": "Opus 4.8",
         "efforts": ["low", "medium", "high", "xhigh", "max"],
         "default_effort": "high",
     },
@@ -40,6 +40,32 @@ def model_supports_effort(model_id: str, effort: str) -> bool:
     return False
 
 
+def _provider_of(model_id: str) -> str | None:
+    provider, _, rest = model_id.partition(":")
+    return provider if rest else None
+
+
+def provider_fallback_pair(model_id: object, effort: object = None) -> tuple[str, str] | None:
+    """Newest supported ``(model_id, effort)`` for the same provider as ``model_id``.
+
+    Keeps a stored selection on its original provider when its exact id has
+    dropped out of the supported set (e.g. an Opus minor-version bump), instead
+    of falling through to the cross-provider global default. Preserves ``effort``
+    when the fallback model supports it, otherwise uses that model's default
+    effort. Returns ``None`` when no supported model shares the provider.
+    """
+    if not isinstance(model_id, str):
+        return None
+    provider = _provider_of(model_id)
+    if provider is None:
+        return None
+    for m in SUPPORTED_MODELS:
+        if _provider_of(m["id"]) == provider:
+            new_effort = effort if (isinstance(effort, str) and effort in m["efforts"]) else None
+            return m["id"], new_effort or m["default_effort"]
+    return None
+
+
 def default_model_pair() -> tuple[str, str]:
     """Hardcoded fallback (model_id, reasoning_effort) used when no team default is set."""
     if DEFAULT_MODEL_ID in SUPPORTED_MODEL_IDS and model_supports_effort(

agent/dashboard/team_settings.py

@@ -14,7 +14,12 @@ from typing import Any, Literal
 from langgraph_sdk import get_client
 from pydantic import BaseModel, model_validator
 
-from .options import SUPPORTED_MODEL_IDS, default_model_pair, model_supports_effort
+from .options import (
+    SUPPORTED_MODEL_IDS,
+    default_model_pair,
+    model_supports_effort,
+    provider_fallback_pair,
+)
 
 logger = logging.getLogger(__name__)
 
@@ -145,10 +150,11 @@ async def get_team_default_model(
 ) -> tuple[str, str]:
     """Return the team-wide default ``(model_id, reasoning_effort)`` for ``role``.
 
-    Always returns a valid pair: the admin-configured pair if set, otherwise the
-    hardcoded fallback from :func:`agent.dashboard.options.default_model_pair`.
-    Invalid stored pairs (unsupported model or mismatched effort) fall back to
-    the hardcoded default rather than propagating bad data.
+    Always returns a valid pair, resolved in order: the admin-configured pair if
+    still supported; otherwise the newest supported model for the same provider
+    (so a stale Anthropic/OpenAI selection stays on its provider rather than
+    jumping cross-provider); otherwise the hardcoded global default from
+    :func:`agent.dashboard.options.default_model_pair`.
     """
     settings = await get_team_settings()
     if role == "agent":
@@ -157,14 +163,7 @@ async def get_team_default_model(
     else:
         model = settings.get("default_reviewer_model")
         effort = settings.get("default_reviewer_reasoning_effort")
-    if (
-        isinstance(model, str)
-        and model in SUPPORTED_MODEL_IDS
-        and isinstance(effort, str)
-        and model_supports_effort(model, effort)
-    ):
-        return model, effort
-    return default_model_pair()
+    return _resolve_default_pair(model, effort)
 
 
 async def get_team_default_subagent_model(
@@ -178,11 +177,19 @@ async def get_team_default_subagent_model(
     else:
         model = settings.get("default_reviewer_subagent_model")
         effort = settings.get("default_reviewer_subagent_reasoning_effort")
+    return _resolve_default_pair(model, effort)
+
+
+def _resolve_default_pair(model: object, effort: object) -> tuple[str, str]:
+    """Supported pair if valid, else same-provider fallback, else global default."""
     if (
         isinstance(model, str)
-        and model in SUPPORTED_MODEL_IDS
         and isinstance(effort, str)
+        and model in SUPPORTED_MODEL_IDS
         and model_supports_effort(model, effort)
     ):
         return model, effort
+    provider_pair = provider_fallback_pair(model, effort)
+    if provider_pair is not None:
+        return provider_pair
     return default_model_pair()

evals/reviewer/config.toml

@@ -5,7 +5,7 @@ max_concurrency = 5
 # Leave blank to use LANGGRAPH_URL or local dev.
 langgraph_url = ""
 assistant_id = "reviewer"
-# models: openai:gpt-5.5, anthropic:claude-opus-4-7, google_genai:gemini-3.5-flash
+# models: openai:gpt-5.5, anthropic:claude-opus-4-8, google_genai:gemini-3.5-flash
 model_id = "google_genai:gemini-3.5-flash"
 reasoning_effort = "medium"
 

tests/test_agent_subagent_models.py

@@ -61,7 +61,7 @@ async def test_agent_uses_profile_subagent_model_override() -> None:
             "agent.server.load_profile",
             new_callable=AsyncMock,
             return_value={
-                "default_model": "anthropic:claude-opus-4-7",
+                "default_model": "anthropic:claude-opus-4-8",
                 "reasoning_effort": "high",
                 "default_subagent_model": "openai:gpt-5.5",
                 "subagent_reasoning_effort": "xhigh",
@@ -81,7 +81,7 @@ async def test_agent_uses_profile_subagent_model_override() -> None:
     assert subagents[0]["model"] is subagent_model
 
     main_call = make_model.call_args_list[0]
-    assert main_call.args == ("anthropic:claude-opus-4-7",)
+    assert main_call.args == ("anthropic:claude-opus-4-8",)
     assert main_call.kwargs["thinking"] == {"type": "adaptive"}
     assert main_call.kwargs["effort"] == "high"
 
@@ -139,7 +139,7 @@ async def test_agent_subagent_inherits_profile_model_override_without_explicit_p
             "agent.server.load_profile",
             new_callable=AsyncMock,
             return_value={
-                "default_model": "anthropic:claude-opus-4-7",
+                "default_model": "anthropic:claude-opus-4-8",
                 "reasoning_effort": "high",
             },
         ),
@@ -153,7 +153,7 @@ async def test_agent_subagent_inherits_profile_model_override_without_explicit_p
     subagents = captured["subagents"]
     assert isinstance(subagents, list)
     assert subagents[0]["model"] is subagent_model
-    assert make_model.call_args_list[0].args == ("anthropic:claude-opus-4-7",)
-    assert make_model.call_args_list[1].args == ("anthropic:claude-opus-4-7",)
+    assert make_model.call_args_list[0].args == ("anthropic:claude-opus-4-8",)
+    assert make_model.call_args_list[1].args == ("anthropic:claude-opus-4-8",)
     assert make_model.call_args_list[1].kwargs["thinking"] == {"type": "adaptive"}
     assert make_model.call_args_list[1].kwargs["effort"] == "high"

tests/test_model_fallback_resolution.py

@@ -0,0 +1,81 @@
+from unittest.mock import AsyncMock, patch
+
+import pytest
+
+from agent.dashboard.agent_overrides import normalize_profile_overrides
+from agent.dashboard.options import (
+    DEFAULT_MODEL_ID,
+    default_model_pair,
+    provider_fallback_pair,
+)
+from agent.dashboard.team_settings import get_team_default_model
+
+STALE_ANTHROPIC = "anthropic:claude-opus-4-7"
+SUPPORTED_ANTHROPIC = "anthropic:claude-opus-4-8"
+
+
+def test_provider_fallback_preserves_provider_and_effort() -> None:
+    assert provider_fallback_pair(STALE_ANTHROPIC, "xhigh") == (SUPPORTED_ANTHROPIC, "xhigh")
+
+
+def test_provider_fallback_uses_default_effort_when_unsupported() -> None:
+    assert provider_fallback_pair(STALE_ANTHROPIC, "bogus") == (SUPPORTED_ANTHROPIC, "high")
+    assert provider_fallback_pair(STALE_ANTHROPIC, None) == (SUPPORTED_ANTHROPIC, "high")
+
+
+def test_provider_fallback_resolves_openai_within_provider() -> None:
+    model, effort = provider_fallback_pair("openai:gpt-5-legacy", "low")
+    assert model.startswith("openai:")
+    assert effort == "low"
+
+
+@pytest.mark.parametrize("model_id", ["unknown:model", "no-colon", "", None, 123])
+def test_provider_fallback_returns_none_without_provider_match(model_id: object) -> None:
+    assert provider_fallback_pair(model_id, "high") is None
+
+
+@pytest.mark.asyncio
+async def test_team_default_stale_anthropic_stays_on_provider() -> None:
+    settings = {
+        "default_agent_model": STALE_ANTHROPIC,
+        "default_agent_reasoning_effort": "xhigh",
+    }
+    with patch(
+        "agent.dashboard.team_settings.get_team_settings",
+        new_callable=AsyncMock,
+        return_value=settings,
+    ):
+        assert await get_team_default_model("agent") == (SUPPORTED_ANTHROPIC, "xhigh")
+
+
+@pytest.mark.asyncio
+async def test_team_default_unknown_provider_falls_back_to_global() -> None:
+    settings = {
+        "default_reviewer_model": "mystery:model",
+        "default_reviewer_reasoning_effort": "high",
+    }
+    with patch(
+        "agent.dashboard.team_settings.get_team_settings",
+        new_callable=AsyncMock,
+        return_value=settings,
+    ):
+        assert await get_team_default_model("reviewer") == default_model_pair()
+
+
+def test_profile_stale_anthropic_upgrades_to_supported() -> None:
+    profile = {"default_model": STALE_ANTHROPIC, "reasoning_effort": "high"}
+    assert normalize_profile_overrides(profile) == (SUPPORTED_ANTHROPIC, "high")
+
+
+def test_profile_without_model_defers_to_team_default() -> None:
+    assert normalize_profile_overrides({"reasoning_effort": "high"}) == (None, None)
+
+
+def test_profile_unknown_provider_defers_to_team_default() -> None:
+    profile = {"default_model": "mystery:model", "reasoning_effort": "high"}
+    assert normalize_profile_overrides(profile) == (None, None)
+
+
+def test_global_default_is_supported() -> None:
+    model, _ = default_model_pair()
+    assert model == DEFAULT_MODEL_ID

tests/test_reviewer.py

@@ -113,7 +113,7 @@ async def test_reviewer_applies_eval_model_and_effort_overrides() -> None:
             "pr_url": "https://github.com/acme/repo/pull/1",
             "base_sha": "base",
             "head_sha": "head",
-            "reviewer_model_id": "anthropic:claude-opus-4-7",
+            "reviewer_model_id": "anthropic:claude-opus-4-8",
             "reviewer_reasoning_effort": "high",
             "reviewer_subagent_model_id": "openai:gpt-5.5",
             "reviewer_subagent_reasoning_effort": "low",
@@ -144,7 +144,7 @@ async def test_reviewer_applies_eval_model_and_effort_overrides() -> None:
         await reviewer.get_reviewer_agent(config)
 
     main_model_call = make_model.call_args_list[0]
-    assert main_model_call.args == ("anthropic:claude-opus-4-7",)
+    assert main_model_call.args == ("anthropic:claude-opus-4-8",)
     assert main_model_call.kwargs["thinking"] == {"type": "adaptive"}
     assert main_model_call.kwargs["effort"] == "high"
     subagent_model_call = make_model.call_args_list[1]
@@ -163,7 +163,7 @@ async def test_reviewer_subagent_inherits_eval_model_without_explicit_override()
             "pr_url": "https://github.com/acme/repo/pull/1",
             "base_sha": "base",
             "head_sha": "head",
-            "reviewer_model_id": "anthropic:claude-opus-4-7",
+            "reviewer_model_id": "anthropic:claude-opus-4-8",
             "reviewer_reasoning_effort": "high",
         },
         "metadata": {},
@@ -192,11 +192,11 @@ async def test_reviewer_subagent_inherits_eval_model_without_explicit_override()
         await reviewer.get_reviewer_agent(config)
 
     main_model_call = make_model.call_args_list[0]
-    assert main_model_call.args == ("anthropic:claude-opus-4-7",)
+    assert main_model_call.args == ("anthropic:claude-opus-4-8",)
     assert main_model_call.kwargs["thinking"] == {"type": "adaptive"}
     assert main_model_call.kwargs["effort"] == "high"
     subagent_model_call = make_model.call_args_list[1]
-    assert subagent_model_call.args == ("anthropic:claude-opus-4-7",)
+    assert subagent_model_call.args == ("anthropic:claude-opus-4-8",)
     assert subagent_model_call.kwargs["thinking"] == {"type": "adaptive"}
     assert subagent_model_call.kwargs["effort"] == "high"
 

tests/test_reviewer_eval_run.py

@@ -14,7 +14,7 @@ def test_reviewer_eval_config_coerces_known_values() -> None:
             "max_concurrency": 2,
             "langgraph_url": "https://example.test",
             "assistant_id": "reviewer",
-            "model_id": "anthropic:claude-opus-4-7",
+            "model_id": "anthropic:claude-opus-4-8",
             "reasoning_effort": "high",
             "score_mode": "surfaced_findings",
             "severity_threshold": "medium",
@@ -29,7 +29,7 @@ def test_reviewer_eval_config_coerces_known_values() -> None:
         "max_concurrency": 2,
         "langgraph_url": "https://example.test",
         "assistant_id": "reviewer",
-        "model_id": "anthropic:claude-opus-4-7",
+        "model_id": "anthropic:claude-opus-4-8",
         "reasoning_effort": "high",
         "score_mode": "surfaced_findings",
         "severity_threshold": "medium",
@@ -43,7 +43,7 @@ def test_reviewer_eval_config_sets_target_env() -> None:
             {
                 "langgraph_url": "https://example.test",
                 "assistant_id": "reviewer",
-                "model_id": "anthropic:claude-opus-4-7",
+                "model_id": "anthropic:claude-opus-4-8",
                 "reasoning_effort": "high",
                 "score_mode": "surfaced_findings",
                 "severity_threshold": "high",
@@ -53,7 +53,7 @@ def test_reviewer_eval_config_sets_target_env() -> None:
 
         assert os.environ["LANGGRAPH_URL"] == "https://example.test"
         assert os.environ["REVIEWER_ASSISTANT_ID"] == "reviewer"
-        assert os.environ["REVIEWER_EVAL_MODEL_ID"] == "anthropic:claude-opus-4-7"
+        assert os.environ["REVIEWER_EVAL_MODEL_ID"] == "anthropic:claude-opus-4-8"
         assert os.environ["REVIEWER_EVAL_REASONING_EFFORT"] == "high"
         assert os.environ["REVIEWER_EVAL_SCORE_MODE"] == "surfaced_findings"
         assert os.environ["REVIEWER_EVAL_SEVERITY_THRESHOLD"] == "high"

tests/test_reviewer_eval_target.py

@@ -29,7 +29,7 @@ def test_eval_target_marks_runs_as_eval_dry_run(monkeypatch: pytest.MonkeyPatch)
 
 
 def test_eval_target_passes_model_overrides(monkeypatch: pytest.MonkeyPatch) -> None:
-    monkeypatch.setenv("REVIEWER_EVAL_MODEL_ID", "anthropic:claude-opus-4-7")
+    monkeypatch.setenv("REVIEWER_EVAL_MODEL_ID", "anthropic:claude-opus-4-8")
     monkeypatch.setenv("REVIEWER_EVAL_REASONING_EFFORT", "high")
 
     configurable = target._build_configurable(
@@ -43,7 +43,7 @@ def test_eval_target_passes_model_overrides(monkeypatch: pytest.MonkeyPatch) ->
         }
     )
 
-    assert configurable["reviewer_model_id"] == "anthropic:claude-opus-4-7"
+    assert configurable["reviewer_model_id"] == "anthropic:claude-opus-4-8"
     assert configurable["reviewer_reasoning_effort"] == "high"