Nov 9, 2026
Meta's Muse Ember evaluated on the Vals Index
We evaluated Meta’s new Muse Ember on Vals Index v2.2 and its components.
-
#12 of 35 on the Vals Index (60.48%) at $0.15 per test, the lowest cost per test of any model on the index.
-
#4 of 36 on the Legal Index (75.69%) and tied for #2 of 36 on HLAB-Vals-1.1 (25.00%). It scores 60.91% on the Finance Index (#10 of 38), 63.83% on Finance Agent v2.1 (#14 of 38) and 83.70% on Vibe Code Bench (#22 of 126).
-
Weaker on Terminal-Bench 4.0 (39.39%, #15 of 62) and Tax Agent Bench (38.23%, #14 of 83).
Muse Ember is a reasoning model with a 1M-token context window, image inputs and tool calling, priced at $0.10 per million input tokens and $0.50 per million output tokens. We ran it at reasoning effort max.
Congrats to the Meta team on the release!