diff --git a/.env.example b/.env.example
index 40e28b9..2d50a2a 100644
--- a/.env.example
+++ b/.env.example
@@ -41,7 +41,7 @@
 # ── Context compression ────────────────────────────────────────────────────────
 CONTEXT_COMPRESSION_ENABLED=true
 CONTEXT_COMPRESSION_THRESHOLD=0.70
-CONTEXT_KEEP_RECENT=8
+CONTEXT_KEEP_RECENT=4
 CONTEXT_SUMMARY_TEMPERATURE=0.3
 CONTEXT_SUMMARY_MAX_TOKENS=3000
 
diff --git a/.env.navi_code.example b/.env.navi_code.example
index 5f4676a..f994898 100644
--- a/.env.navi_code.example
+++ b/.env.navi_code.example
@@ -41,7 +41,7 @@
 # ── Context compression ────────────────────────────────────────────────────────
 CONTEXT_COMPRESSION_ENABLED=true
 CONTEXT_COMPRESSION_THRESHOLD=0.70
-CONTEXT_KEEP_RECENT=8
+CONTEXT_KEEP_RECENT=4
 CONTEXT_SUMMARY_TEMPERATURE=0.3
 CONTEXT_SUMMARY_MAX_TOKENS=3000
 
diff --git a/docs/api.md b/docs/api.md
index 8dff38e..7ea0277 100644
--- a/docs/api.md
+++ b/docs/api.md
@@ -503,7 +503,7 @@
   "compression": {
     "enabled": true,
     "threshold": 0.9,
-    "keep_recent": 8,
+    "keep_recent": 4,
     "summary_max_tokens": 6000
   },
   "llm": {
diff --git a/docs/config.md b/docs/config.md
index 18ac2c5..26dcf2b 100644
--- a/docs/config.md
+++ b/docs/config.md
@@ -247,7 +247,7 @@
 |---|---|---|---|
 | `CONTEXT_COMPRESSION_ENABLED` | bool | `true` | Enable/disable automatic context compression |
 | `CONTEXT_COMPRESSION_THRESHOLD` | float | `0.90` | Trigger compression at this fraction of `OLLAMA_NUM_CTX` |
-| `CONTEXT_KEEP_RECENT` | int | `8` | Number of recent conversation turns to keep verbatim |
+| `CONTEXT_KEEP_RECENT` | int | `4` | Number of recent conversation turns to keep verbatim |
 | `CONTEXT_SUMMARY_TEMPERATURE` | float | `0.3` | Temperature for the summarization LLM call |
 | `CONTEXT_SUMMARY_MAX_TOKENS` | int | `6000` | Max output tokens for the summary LLM call |
 | `OUTPUT_RESERVE_TOKENS` | int | `2048` | Headroom reserved for model response in context size checks |
@@ -371,7 +371,7 @@
 # Context compression
 CONTEXT_COMPRESSION_ENABLED=true
 CONTEXT_COMPRESSION_THRESHOLD=0.90
-CONTEXT_KEEP_RECENT=8
+CONTEXT_KEEP_RECENT=4
 
 # Persona
 NAVI_PERSONA_FILE=persona.txt
diff --git a/docs/sessions.md b/docs/sessions.md
index d6ecc73..3d4500a 100644
--- a/docs/sessions.md
+++ b/docs/sessions.md
@@ -122,7 +122,7 @@
 - `context_compression_enabled: bool = True`
 - `context_compression_threshold: float = 0.90` — trigger at 90% of the active window (real tokens)
 - `context_compression_target: float = 0.65` — hysteresis target: after compression the kept region should fit ~65% of the active window, so the trigger doesn't re-fire a few messages later
-- `context_keep_recent: int = 8` — keep last N conversational turns verbatim
+- `context_keep_recent: int = 4` — keep last N conversational turns verbatim
 - `context_summary_temperature: float = 0.3`
 - `context_summary_max_tokens: int = 6000` — max output tokens for the summary LLM call
 - `output_reserve_tokens: int = 2048` — headroom reserved for the response in `check_context_size`
diff --git a/navi/config.py b/navi/config.py
index ca8744a..9442840 100644
--- a/navi/config.py
+++ b/navi/config.py
@@ -223,7 +223,7 @@
     context_compression_enabled: bool = True
     context_compression_threshold: float = 0.90   # trigger at 90% of ollama_num_ctx
     context_compression_target: float = 0.65       # hysteresis target — after compression the kept region should fit ~65% of the window (trigger 90% → target 65% leaves real headroom instead of skimming just below the trigger and re-firing a few messages later)
-    context_keep_recent: int = 8                   # conversational turns to keep verbatim
+    context_keep_recent: int = 4                   # conversational turns to keep verbatim
     context_summary_temperature: float = 0.3
     context_summary_max_tokens: int = 6000         # max output tokens for the summary LLM call
     output_reserve_tokens: int = 2048              # headroom reserved for model response in context checks
diff --git a/tests/unit/api/test_session_info.py b/tests/unit/api/test_session_info.py
index 935089d..acba3d9 100644
--- a/tests/unit/api/test_session_info.py
+++ b/tests/unit/api/test_session_info.py
@@ -197,7 +197,7 @@
         assert info["context_token_count"] == 51013
         assert info["archive_threshold"] == 2050
         assert info["compression"]["enabled"] is True
-        assert info["compression"]["keep_recent"] == 8
+        assert info["compression"]["keep_recent"] == 4
 
     async def test_a_profile_override_replaces_the_compression_default(self):
         info = await _call(profile=_profile(compression_keep_recent=3,
diff --git a/tests/unit/core/test_agent.py b/tests/unit/core/test_agent.py
index 8e89079..4f56c3c 100644
--- a/tests/unit/core/test_agent.py
+++ b/tests/unit/core/test_agent.py
@@ -646,7 +646,7 @@
     @pytest_asyncio.fixture
     async def session_with_history(self, agent, session):
         """A session with enough turns that there is something to summarize."""
-        # context_keep_recent defaults to 8 turns; add 10 so the oldest turns
+        # context_keep_recent defaults to 4 turns; add 10 so the oldest turns
         # fall into to_summarize (>= 2 messages) and compress_context does work.
         for i in range(10):
             session.context.append(Message(role="user", content=f"user {i}"))