Microsoft MAI claims SOTA 3T-param in-house base models, post-training results coming
soldni · x · 2026-10-16
Mustafa Suleyman announced MAI's state-of-the-art pre-trained base models in the 3T-parameter weight class, built entirely in-house at Microsoft, with post-training results "coming soon." A researcher added that postraining simply takes time.
Benchmark charts show MAI-Base-2 (80B/2.8T) beating MiMo-V2-6-Pro-RL, Qwen3.8, DeepSeek-V4-Base and Kimi-K3 on held-out code, QA, STEM and math evals (bits per byte, lower is better) — e.g. QA 0.4162 vs 0.44+, Math 0.3561 vs 0.38+. Even the smaller MAI-Base-1 (35B/1.1T) outperforms most larger rivals.
Related event: Microsoft Unveils MAI-Base-2, 2.8T-Parameter MoE Base Model(3 posts)→
More from Models
- Meta officially launches Muse Spark Open: 816B total, 42B active, 1M context — airesearch12 · 2026-10-16
- OpenAI simplifies API usage tiers into Build/Launch/Grow with automatic upgrades at $5/$100/$500 — prd_008 · 2026-10-16
- Xiaomi's MiMo RL dashboard reveals $3.47M spent on just 30 training steps — bycloud · 2026-10-16
- Analyst: in a year, models 2-3x better across the board at 10x lower cost, token use up 10-20x — DeryaTR_ · 2026-10-16
- Google revamps Gemma docs, site teases Gemma 4 open models — rseroter · 2026-10-16
- Respan's Span-02 brings anytime reasoning: interrupt it mid-thought and still get the best answer — techNmak · 2026-10-16