fix: sync tier 4 default threshold to 50000 tokens (was stale at 4000)
This commit is contained in:
+1
-1
@@ -188,7 +188,7 @@ def route(rd, tier):
|
||||
if result: return result
|
||||
|
||||
# TIER 4: Default — MoE first, VLM helps, Dense last (slow)
|
||||
if t <= 4000:
|
||||
if t <= 50000:
|
||||
candidates = [m for m in ["qwen3.6-35B-A3B","qwen3.5-9b-vlm","qwen3.6-27B-code"] if m in avail]
|
||||
result = select_best_gpu(candidates, "default")
|
||||
if result: return result
|
||||
|
||||
Reference in New Issue
Block a user