Microsoft trains MAI-Base-2, its most efficient frontier foundation model yet

What you could do with it

• An enterprise could license MAI-Base-2's model weights through Microsoft's Foundry platform to power its own AI services ¹.
• A researcher studying AI scaling could follow Microsoft's 'source scaling' and 'scaling ladder' methods to plan their own experiments ¹.

A faster way to train giant AI

Microsoft has finished pre-training a new frontier foundation model called MAI-Base-2. The model belongs to a family that also includes MAI-Base-1 and MAI-Thinking-1, the reasoning-focused version Microsoft announced in June. MAI-Base-2 is built on a Mixture of Experts (MoE) design, a technique that spreads a massive model across many smaller "expert" networks so only a subset activates for each task.

MAI-Base-2 has 2.8 trillion total parameters but uses just 80 billion active ones during inference. That sparse approach means it can handle complex reasoning without paying the full computational cost of its enormous size.

Efficiency that scales

On held-out evaluations, Microsoft says MAI-Base-2 outperforms every open-weight model it tested. More striking is its efficiency: it requires 4 to 12 times fewer pre-training FLOPs (floating-point operations, the standard unit for measuring AI training compute) than DeepSeek-V4-Pro-Base, the strongest open base model Microsoft tested.

Microsoft is treating that efficiency gain not as an end in itself but as a resource to reinvest. The company says it has put the saved compute into training a larger model, suggesting future iterations will build directly on these gains.

Where it fits in the frontier

Frontier AI refers to the most capable models at the cutting edge of research, often developed by well-funded labs with access to enormous compute clusters. Microsoft's decision to share the training approach, even if it keeps the model weights private, fits a growing pattern of companies using published research to guide the field.

MAI-Base-2 slots into Microsoft's broader MAI model family. Earlier this year the company introduced MAI-Base-1, and in June released MAI-Thinking-1 as a post-trained reasoning layer on top of that base. Together the series covers both general-purpose and reasoning-optimized paths, giving developers a choice depending on the task.

What comes next

No public release date or access details for MAI-Base-2 have been announced. Microsoft has not disclosed whether the model will eventually become open-weight or remain proprietary. What is clear is that the efficiency improvements are measurable and substantial enough to reshape how the company plans its next training run.

How good is it

On held-out evaluations, MAI-Base-2 outperforms every open-weight model Microsoft tested, including DeepSeek-V4-Pro-Base. Microsoft also reports results for post-trained checkpoints Kimi K3, Qwen3.8, and MiMo-V2.6-Pro-RL, noting that post-training may affect these evaluations. ¹. It is 4 to 12 times more FLOPs-efficient than DeepSeek-V4-Pro-Base, meaning it reaches better results while using far less computing work to train . For held-out text, MAI-Base-2 achieves a lower bits-per-byte score than the other models, which indicates better language understanding . Since MAI-Base-1, Microsoft's training efficiency has improved by 6.3 to 12.7 times .

Corrections and updates

• Corrected on 23 October 2026. An earlier version said MAI-Base-2 was like its predecessor in using MoE, but the source did not confirm MAI-Base-1 was MoE. It also said MAI-Base-2 beat every open-weight model in its class, but the source specified only the models Microsoft tested.

Key facts

• MAI-Base-2 is an MoE model with 2.8 trillion total and 80 billion active parameters.
• It outperforms every open-weight model Microsoft tested on held-out evaluations.
• It is 4-12 times more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base.
• Microsoft is reinvesting compute savings into training a larger follow-on model.
• The model is part of the MAI family alongside MAI-Base-1 and MAI-Thinking-1.

Sources

1. Frontier Pre-training at Microsoft AI | Microsoft AI — microsoft.ai