Flashinfer MOE FP8 support for Mistral Large 3. (#15422)

Co-authored-by: elvischenv <219235043+elvischenv@users.noreply.github.com>

test/registered/8-gpu-models/test_mistral_large3.py

@@ -46,6 +46,7 @@ class TestMistralLarge3(unittest.TestCase):
         base_args = [
             "--tp=8",
             "--attention-backend=trtllm_mla",
+            "--moe-runner-backend=flashinfer_trtllm",
             "--model-loader-extra-config",
             '{"enable_multithread_load": true}',
             "--chat-template=mistral",
@@ -58,10 +59,6 @@ class TestMistralLarge3(unittest.TestCase):
             "--speculative-num-draft-tokens=4",
             "--kv-cache-dtype=auto",
         ]
-        # TODO: add this to base args when FP8 TRTLLM moe is supported
-        nvfp4_args = [
-            "--moe-runner-backend=flashinfer_trtllm",
-        ]
 
         variants = [
             # Variant: "basic" - FP8 model + TP=8 + trtllm_mla backend
@@ -83,7 +80,7 @@ class TestMistralLarge3(unittest.TestCase):
             ModelLaunchSettings(
                 MISTRAL_LARGE3_NVFP4_MODEL_PATH,
                 tp_size=8,
-                extra_args=base_args + nvfp4_args,
+                extra_args=base_args,
                 variant="NVFP4",
             ),
         ]