Terminal-Bench 2.1 ↗67.79 %
MiMo V2.6 Pro (Terminus 2) · Terminal-Bench 2.1 · Score ↗
Xiaomi / MODEL EVIDENCE
MiMo V2.6 Pro (Terminus 2)
1 benchmarks · 1 observations · 1 sources
Scores at a glance
1 benchmarks · large dot: MiMo V2.6 Pro (Terminus 2); small dots: closest models where reported
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
1 of 1 shared benchmark comparisons
1 of 1 shared benchmark comparisons
1 of 1 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
POST-TRAINING / SOURCE DISCLOSURE
What the MiMo-V2.6 Pro report says
Xiaomi describes large-scale multitask reinforcement learning with asynchronous generation, varied agent harnesses and group-based grading. Freezing the MoE router is one of the stated stability measures.
- Multitask reinforcement learning
- Asynchronous rollout infrastructure
- Multiple agent harnesses and group-based grading
- Frozen MoE router during the reported scale-up
The reviewed announcement describes the scale-up and stability mechanisms. Frozen routing is not the same as routing replay, and no specific policy loss is inferred from the presence of group grading.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
Names used in source tables
MiMo V2.6 Pro (Terminus 2)