diff --git a/hindsight-api-slim/hindsight_api/engine/memory_engine.py b/hindsight-api-slim/hindsight_api/engine/memory_engine.py
index 4f633e91b..8eb9830e9 100644
--- a/hindsight-api-slim/hindsight_api/engine/memory_engine.py
+++ b/hindsight-api-slim/hindsight_api/engine/memory_engine.py
@@ -52,6 +52,7 @@
 # Context variable for current schema (async-safe, per-task isolation)
 # Note: default is None, actual default comes from config via get_current_schema()
 _current_schema: contextvars.ContextVar[str | None] = contextvars.ContextVar("current_schema", default=None)
+MENTAL_MODEL_PENDING_CONTENT = "Generating content..."
 
 
 def get_current_schema() -> str:
@@ -7596,7 +7597,8 @@ async def refresh_mental_model(
             # stub out the DB don't hit an unexpected pool access).
             use_delta = False
             stored_structured_content: dict[str, Any] | None = None
-            if refresh_mode == "delta" and current_content:
+            has_delta_baseline = bool(current_content) and current_content != MENTAL_MODEL_PENDING_CONTENT
+            if refresh_mode == "delta" and has_delta_baseline:
                 backend = await self._get_backend()
                 async with acquire_with_retry(backend) as conn:
                     tracking_row = await conn.fetchrow(
diff --git a/hindsight-api-slim/hindsight_api/engine/providers/openai_compatible_llm.py b/hindsight-api-slim/hindsight_api/engine/providers/openai_compatible_llm.py
index 9bb6fdba0..0f02949d5 100644
--- a/hindsight-api-slim/hindsight_api/engine/providers/openai_compatible_llm.py
+++ b/hindsight-api-slim/hindsight_api/engine/providers/openai_compatible_llm.py
@@ -306,15 +306,19 @@ def __init__(
             self.api_key = "local"
 
         # Validate API key for cloud providers
-        if self.provider in (
-            "openai",
-            "groq",
-            "minimax",
-            "deepseek",
-            "openrouter",
-            "zai",
-            "opencode-go",
-        ) and not self.api_key:
+        if (
+            self.provider
+            in (
+                "openai",
+                "groq",
+                "minimax",
+                "deepseek",
+                "openrouter",
+                "zai",
+                "opencode-go",
+            )
+            and not self.api_key
+        ):
             raise ValueError(f"API key is required for {self.provider}")
 
         # Service tier configuration (from config, not env vars)
diff --git a/hindsight-api-slim/tests/test_mental_model_delta.py b/hindsight-api-slim/tests/test_mental_model_delta.py
index fbbb41952..4f1a18214 100644
--- a/hindsight-api-slim/tests/test_mental_model_delta.py
+++ b/hindsight-api-slim/tests/test_mental_model_delta.py
@@ -187,6 +187,46 @@ async def test_delta_mode_empty_content_falls_back_to_full(
 
         await memory.delete_bank(bank_id, request_context=request_context)
 
+    async def test_delta_mode_pending_placeholder_falls_back_to_full(
+        self,
+        memory: MemoryEngine,
+        request_context: RequestContext,
+        patch_reflect,
+        patch_llm_call,
+    ):
+        """The async creation placeholder is not a real delta baseline.
+
+        A first refresh for a newly-created model must do a full recall over
+        pre-existing facts instead of scoping recall to last_refreshed_at.
+        """
+        bank_id = f"test-delta-placeholder-{uuid.uuid4().hex[:8]}"
+        await memory.get_bank_profile(bank_id, request_context=request_context)
+
+        mm = await memory.create_mental_model(
+            bank_id=bank_id,
+            name="Backend Overview",
+            source_query="What is the backend architecture?",
+            content="Generating content...",
+            trigger={"mode": "delta"},
+            request_context=request_context,
+        )
+
+        reflect_calls = patch_reflect(memory, text="# Backend\n\nFull fresh synthesis.")
+        llm_calls = patch_llm_call(memory, returns="should-not-be-called")
+
+        refreshed = await memory.refresh_mental_model(
+            bank_id=bank_id, mental_model_id=mm["id"], request_context=request_context
+        )
+
+        assert refreshed["content"] == "# Backend\n\nFull fresh synthesis."
+        assert len(llm_calls) == 0
+        assert "created_after" not in reflect_calls[0]
+        rr = refreshed.get("reflect_response") or {}
+        assert rr.get("delta_applied") is not True
+        assert rr.get("delta_skipped_reason") is None
+
+        await memory.delete_bank(bank_id, request_context=request_context)
+
     async def test_delta_mode_source_query_change_falls_back_to_full(
         self,
         memory: MemoryEngine,
diff --git a/skills/hindsight-docs/references/developer/models.md b/skills/hindsight-docs/references/developer/models.md
index b50e625a4..51278726a 100644
--- a/skills/hindsight-docs/references/developer/models.md
+++ b/skills/hindsight-docs/references/developer/models.md
@@ -30,6 +30,7 @@ Used for fact extraction, entity resolution, mental model consolidation, and ans
 - MiniMax
 - DeepSeek
 - z.ai
+- opencode-go
 - Volcano Engine
 - OpenRouter
 - OpenAI Codex
diff --git a/skills/hindsight-docs/references/faq.md b/skills/hindsight-docs/references/faq.md
index 8b8b96a62..f4f8f2aca 100644
--- a/skills/hindsight-docs/references/faq.md
+++ b/skills/hindsight-docs/references/faq.md
@@ -76,6 +76,7 @@ Browse all supported integrations in the Integrations Hub.
 - MiniMax
 - DeepSeek
 - z.ai
+- opencode-go
 - Volcano Engine
 - OpenRouter
 - OpenAI Codex