HUMAIN M3
HUMAIN
- Overall
- 67.4%
- Arabic
- 59.0%
- Saudi
- 63.8%
Direct model calls · default thinking
Historical acceptable answer rate
Model catalog
Explore models
Browse models evaluated on Arabic and Saudi tasks. Inspect a model’s results and specifications, or use Leaderboards to compare task rankings.
MiniMax M3
MiniMax
- Overall
- 66.1%
- Arabic
- 55.4%
- Saudi
- 65.8%
Direct model calls · default thinking
Historical acceptable answer rate
Historical evaluation dates and publication dates have not been verified. Read the evidence and scoring policy →
Operational metrics
Partial execution-time samples are available under Inference. Controlled speed, token throughput, and cost are not established and are not used to rank these models.