diff --git a/agentd/core.py b/agentd/core.py
index 1c1164e..27a3fa8 100644
--- a/agentd/core.py
+++ b/agentd/core.py
@@ -491,6 +491,8 @@ class Agentd:
                 r["model"] = label[6:]
             elif label.startswith("effort:"):
                 r["effort"] = label[7:]
+            elif label.startswith("tier:"):
+                r["tier"] = label[5:]
             elif label.startswith("cap:"):
                 caps.append(label[4:])
             elif parse_budget(label) is not None:
@@ -503,6 +505,10 @@ class Agentd:
         ecfg = self.cfg.executors[r["executor"]]
         r["model"] = r["model"] or (ecfg.get("models") or [None])[0]
         r["effort"] = r["effort"] or ecfg.get("default_effort")
+        if r["tier"] is None:  # per-model default unless a label set one; `tier:none` = no tier argument
+            r["tier"] = (ecfg.get("tier_defaults") or {}).get(r["model"])
+        if r["tier"] in ("none", ""):
+            r["tier"] = None
         return r
 
     def protocol(self, pcfg):
diff --git a/config/agentd.example.yaml b/config/agentd.example.yaml
index dac154c..2afbd06 100644
--- a/config/agentd.example.yaml
+++ b/config/agentd.example.yaml
@@ -45,6 +45,7 @@ executors:
     run: "codex exec --json -C '{repo}' -s workspace-write -m {model} -c 'model_reasoning_effort=\"{effort}\"' {tier} -o '{out}' -"
     resume: "codex exec resume {thread} --json -m {model} -c 'model_reasoning_effort=\"{effort}\"' {tier} -o '{out}' -"
     tier_arg: "-c 'service_tier=\"{tier}\"'"
+    tier_defaults: {gpt-6.1-sol: fast}   # tier when no label sets one (`tier:<x>`, `tier:none` = no tier)
     thread_from: events.jsonl:thread.started.thread_id
     last_message: '{out}'
     models: [gpt-6.1-sol, gpt-6-astra]
diff --git a/config/agentd.yaml b/config/agentd.yaml
index 2e7aa15..4ca7f1c 100644
--- a/config/agentd.yaml
+++ b/config/agentd.yaml
@@ -44,6 +44,7 @@ executors:
     # `exec resume` takes no -C/-s: the runner already cd's into the repo, the sandbox goes in as a -c override
     resume: "codex exec resume {thread} --json -m {model} -c 'model_reasoning_effort=\"{effort}\"' -c 'sandbox_mode=\"workspace-write\"' {tier} -o '{out}' -"
     tier_arg: "-c 'service_tier=\"{tier}\"'"
+    tier_defaults: {gpt-6.1-sol: fast}   # tier when no label sets one (`tier:<x>`, `tier:none` = no tier)
     thread_from: events.jsonl:thread.started.thread_id
     last_message: '{out}'
     models: [gpt-6.1-sol, gpt-6-astra]
diff --git a/config/schema.json b/config/schema.json
index ba0250f..5409259 100644
--- a/config/schema.json
+++ b/config/schema.json
@@ -76,6 +76,7 @@
         "default_effort": {"type": "string"},
         "permission_mode": {"type": "string"},
         "tier_arg": {"type": "string"},
+        "tier_defaults": {"type": "object", "additionalProperties": {"type": "string"}},
         "script": {"type": "string"}
       }
     },
diff --git a/tests/fakes.py b/tests/fakes.py
index 3e3a8c1..57b1a1e 100644
--- a/tests/fakes.py
+++ b/tests/fakes.py
@@ -9,7 +9,8 @@ from agentd.store import Store
 
 STATES = {"Backlog": "s-bl", "Todo": "s-todo", "In Progress": "s-run", "Waiting": "s-wait", "Review": "s-rev", "Done": "s-done"}
 MEMBERS = {"window10": "u-w10", "oregon": "u-or", "kappiee": "u-k"}
-LABELS = {"model:astra": "l-astra", "needs-game": "l-game", "exec:claude": "l-claude", "budget:30m": "l-b30", "cap:gpu": "l-gpu"}
+LABELS = {"model:astra": "l-astra", "needs-game": "l-game", "exec:claude": "l-claude", "budget:30m": "l-b30", "cap:gpu": "l-gpu",
+          "tier:none": "l-tnone", "tier:flex": "l-tflex"}
 
 
 def config_data():
diff --git a/tests/test_core.py b/tests/test_core.py
index afd285d..e1d3873 100644
--- a/tests/test_core.py
+++ b/tests/test_core.py
@@ -362,5 +362,32 @@ class EnvFailureTest(unittest.TestCase):
         h.close()
 
 
+
+class TierDefaultTest(unittest.TestCase):
+    """executors.<x>.tier_defaults: tier per model unless a label (`tier:<x>`, `tier:none`) sets one."""
+    def run_with(self, labels):
+        data = config_data()
+        data["executors"]["codex"]["tier_defaults"] = {"sol": "fast"}
+        h = Harness(data)
+        try:
+            iid = h.plane.add_issue(1, labels=labels)
+            h.tick()
+            return h.execs[("win10", "codex")].started[0].tier, h.plane.texts(iid)[-1]
+        finally:
+            h.close()
+
+    def test_default_model_gets_its_tier(self):
+        tier, text = self.run_with(())
+        self.assertEqual(tier, "fast")
+        self.assertIn("codex / sol / medium / fast @ win10", text)
+
+    def test_other_model_has_no_default(self):
+        self.assertIsNone(self.run_with(("model:astra",))[0])
+
+    def test_labels_override(self):
+        self.assertEqual(self.run_with(("tier:flex",))[0], "flex")
+        self.assertIsNone(self.run_with(("tier:none",))[0])
+
+
 if __name__ == "__main__":
     unittest.main()