diff --git a/strix/config/settings.py b/strix/config/settings.py
index e9ed84272..f6c85dd03 100644
--- a/strix/config/settings.py
+++ b/strix/config/settings.py
@@ -59,9 +59,9 @@ class LlmSettings(BaseSettings):
         alias="STRIX_PROMPT_CACHE",
     )
     # Providers cache prompts in fixed-size token blocks, so a fully cached prompt
-    # reads back rounded down to a multiple of this. 64 is what the GLM calls in
-    # local runs showed; it's a per-deployment setting (vLLM defaults to 16).
-    cache_block_tokens: int = Field(default=64, ge=1, alias="STRIX_CACHE_BLOCK_TOKENS")
+    # can read back up to a block short. 128 covers the largest common size
+    # (OpenAI; DeepSeek and GLM use 64, vLLM defaults to 16).
+    cache_block_tokens: int = Field(default=128, ge=1, alias="STRIX_CACHE_BLOCK_TOKENS")
     disable_streaming: bool = Field(
         default=False,
         alias="LLM_DISABLE_STREAMING",
diff --git a/strix/report/usage.py b/strix/report/usage.py
index f84d02e8e..bb20b254d 100644
--- a/strix/report/usage.py
+++ b/strix/report/usage.py
@@ -99,12 +99,11 @@ class LLMUsageLedger:
         tally.cached_tokens += cached_tokens
         if agent_id:
             previous = self._last_input_tokens.get(agent_id, 0)
-            # The previous prompt is a prefix of this one, so every full block of it
-            # should read back cached. A shrinking prompt means compaction rewrote
-            # it, so a miss is expected.
-            expected = previous - (previous % cache_block_tokens)
-            missed = expected - cached_tokens
-            if input_tokens >= previous and missed > 0:
+            # The previous prompt is a prefix of this one, so all of it but a
+            # partial last block should read back cached. A shrinking prompt means
+            # compaction rewrote it, so a miss is expected.
+            missed = previous - cached_tokens
+            if input_tokens >= previous and missed >= cache_block_tokens:
                 tally.cache_misses += 1
                 tally.missed_tokens += missed
             self._last_input_tokens[agent_id] = input_tokens
diff --git a/tests/test_cost_tracking.py b/tests/test_cost_tracking.py
index c9bc283b5..4c0ac9223 100644
--- a/tests/test_cost_tracking.py
+++ b/tests/test_cost_tracking.py
@@ -307,7 +307,7 @@ def test_provider_tally_survives_run_record_round_trip() -> None:
             input_tokens=input_tokens,
             cached_tokens=cached_tokens,
             cost=cost,
-            cache_block_tokens=64,
+            cache_block_tokens=128,
         )
 
     restored = LLMUsageLedger()
@@ -329,8 +329,8 @@ def test_provider_tally_counts_cache_misses_per_agent() -> None:
     ledger = LLMUsageLedger()
     calls = [
         ("Z.AI", "a1", 1000, 0),  # first call: nothing to miss
-        ("Z.AI", "a1", 1200, 960),  # previous 1000 cached, rounded down to 64s
-        ("DeepInfra", "a1", 1500, 200),  # 1152 of the previous 1200 due, 952 lost
+        ("Z.AI", "a1", 1200, 960),  # 40 short of the previous 1000: within a block
+        ("DeepInfra", "a1", 1500, 200),  # 1000 of the previous 1200 lost
         ("Z.AI", "a2", 800, 0),  # another agent's first call
         ("Z.AI", "a1", 600, 0),  # prompt shrank: compaction, not a miss
     ]
@@ -341,12 +341,12 @@ def test_provider_tally_counts_cache_misses_per_agent() -> None:
             input_tokens=input_tokens,
             cached_tokens=cached_tokens,
             cost=0.0,
-            cache_block_tokens=64,
+            cache_block_tokens=128,
         )
 
     providers = ledger.to_record()["providers"]
     assert providers["DeepInfra"]["cache_misses"] == 1
-    assert providers["DeepInfra"]["missed_tokens"] == 952
+    assert providers["DeepInfra"]["missed_tokens"] == 1000
     assert providers["Z.AI"]["cache_misses"] == 0