Update FLUX 3 Action model card

README.md

@@ -16,17 +16,17 @@ FLUX 3 Action is an open weights 7B world action model. It takes camera frames,
 
 For more information, read the [documentation](https://docs.bfl.ai/flux_3/flux3_action_overview).
 
-Fine-tuned on [DROID](https://droid-dataset.github.io/), FLUX 3 Action places first on the RoboLab-120 benchmark at 42.6% task success.
+Fine-tuned on [DROID](https://droid-dataset.github.io/), FLUX 3 Action places first on the RoboLab-120 benchmark at 42.92% task success.
 
 ## Evaluation
 
-RoboLab-120 is 120 tabletop tasks in Isaac Sim, 10 trials each, on a DROID-style Franka setup; a trial succeeds only if the task is completed as instructed. The full board is on the [RoboLab leaderboard](https://research.nvidia.com/labs/srl/projects/robolab/leaderboard.html). Latencies were measured on one H200.
+RoboLab-120 is 120 tabletop tasks in Isaac Sim, 10 trials each, on a DROID-style Franka setup; a trial succeeds only if the task is completed as instructed. The full board is on the [RoboLab leaderboard](https://research.nvidia.com/labs/srl/projects/robolab/leaderboard.html).
 
-| Model | Type | Success | Parameters | Plan latency |
-| --- | --- | --- | --- | --- |
-| FLUX 3 Action | WAM | 42.6% | 7B | 481 ms bf16, 290 ms FP8 |
-| Cosmos3-Nano-Policy | WAM | 36.8% | 16B | 724 ms bf16 |
-| π0.5 | VLA | 28.0% | 3.3B | |
+| Model | Type | Success | Parameters |
+| --- | --- | --- | --- |
+| FLUX 3 Action | WAM | 42.92% | 7B |
+| Cosmos3-Nano-Policy | WAM | 36.8% | 16B |
+| π0.5 | VLA | 28.0% | 3.3B |
 
 WAM: predicts future frames and actions together. VLA: a vision-language model that outputs actions directly.