---
type: "post"
id: "microsoft-ai-says-it-has-completed-pretraining-mai-9757e7"
slug: "microsoft-ai-says-it-has-completed-pretraining-mai-9757e7"
title: "Microsoft AI says it has completed pretraining MAI-Base-2"
description: "Microsoft’s lab note reports completion of pretraining for MAI-Base-2, an MoE foundation model with 80 billion active and 2.8 trillion total parameters."
retrieval_nugget: "Microsoft’s lab note reports completion of pretraining for MAI-Base-2, an MoE foundation model with 80 billion active and 2.8 trillion total parameters."
published_at: "2026-10-16"
updated_at: "2026-10-16"
record_date: "2026-10-16"
date_kind: "published_at"
topics: ["models"]
source_urls: ["https://microsoft.ai/?post_type=new"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:microsoft-ai-says-it-has-completed-pretraining-mai-9757e7"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:addb1935d3341956b0608d828f93eec7597ebd4f5699ec0d2ac2ccd2fcc0021a","reviewed_at":"2026-10-16T04:04:01.495Z","source_evidence_count":1,"verified_claim_count":2,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":2,"implication_count":1,"watch_condition_count":2,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"the announcement is a compute-efficiency claim tied to a completed pretraining run, not evidence of model availability or an independently established performance lead.","observed_facts":[{"text":"Microsoft says MAI-Base-2 is an MoE foundation model with 80 billion active and 2.8 trillion total parameters.","source_urls":["https://microsoft.ai/?post_type=new"]},{"text":"Microsoft reports 4–12x pretraining-FLOPs efficiency versus the strongest open base model it tested, DeepSeek-V4-Pro-Base, and says it used the savings to train a larger model.","source_urls":["https://microsoft.ai/?post_type=new"]}],"mechanism":"Improved training compute efficiency can be reinvested in model scale rather than treated only as a cost reduction.","why_now":"Microsoft has publicly marked a completed run and attached a quantified comparison, making the efficiency claim a concrete research milestone.","implications":["The claimed efficiency could let Microsoft pursue larger training runs within a given compute budget, if the comparison holds beyond its own evaluations."],"evidence_boundary":"The source is Microsoft’s account of its own tests; it does not supply independent validation here, establish access or release timing, or substantiate MAI-Base-3 training preparations.","watch_conditions":["Whether Microsoft publishes evaluation details or an independently reproducible comparison.","Release/access information and evidence clarifying the comparison methodology."],"related_records":[],"new_branch_reason":"No prior New Runtime record is supplied, so this starts a distinct analytical record for this specific completed training run and its efficiency claim, without asserting a link to an earlier record."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/microsoft-ai-says-it-has-completed-pretraining-mai-9757e7/","markdown":"https://newruntime.com/posts/microsoft-ai-says-it-has-completed-pretraining-mai-9757e7.md","json":"https://newruntime.com/posts/microsoft-ai-says-it-has-completed-pretraining-mai-9757e7.json"}
---

# Microsoft AI says it has completed pretraining MAI-Base-2

## Retrieval answer

Microsoft’s lab note reports completion of pretraining for MAI-Base-2, an MoE foundation model with 80 billion active and 2.8 trillion total parameters.

Microsoft’s lab note reports completion of pretraining for MAI-Base-2, an MoE foundation model with 80 billion active and 2.8 trillion total parameters. The exact thesis of this note is: the announcement is a compute-efficiency claim tied to a completed pretraining run, not evidence of model availability or an independently established performance lead.

Observed fact: Microsoft says MAI-Base-2 outperformed every open-weight model it tested on held-out evaluations. Observed fact: it reports 4–12x greater pretraining-FLOPs efficiency than the strongest open base model it tested, DeepSeek-V4-Pro-Base, and says it put the savings into training a larger model. Improved training compute efficiency can be reinvested in model scale rather than treated only as a cost reduction. Microsoft has publicly marked a completed run and attached a quantified comparison, making the efficiency claim a concrete research milestone.

The claimed efficiency could let Microsoft pursue larger training runs within a given compute budget, if the comparison holds beyond its own evaluations. The source is Microsoft’s account of its own tests; it does not supply independent validation here, establish access or release timing, or substantiate MAI-Base-3 training preparations. Whether Microsoft publishes evaluation details or an independently reproducible comparison. Further signals to watch include release/access information and evidence clarifying the comparison methodology. Since no prior New Runtime record is supplied, this starts a distinct analytical record for this specific completed training run and its efficiency claim, without asserting a link to an earlier record.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Models](https://newruntime.com/topics/models.md): Explore the models topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)