Updated daily at 05:00 UTC
Models
AI models
1521 current models, newest first. Each links to its benchmark results, with who ran them.
Best by specialty
Independent and third-party results only · method category-2026-09-29.1
Not ranked yet
Too few independent or third-party results to rank:
Scientific reasoning, Mathematics, Long context, OCR & documents, Law, Medicine & health, Finance, Multilingual, Instruction following, Writing, Speech & audio.
All current models
Best for
- Knowledge & reasoning
- Scientific reasoning
- Mathematics
- Coding & software engineering
- Agentic tool use
- Computer & browser use
- Professional knowledge work
- Long context
- Vision & multimodal understanding
- OCR & documents
- Law
- Medicine & health
- Finance
- Multilingual
- Factuality & hallucination
- Instruction following
- Writing
- Speech & audio
- Human preference
Kind
Context
Input price
Weights· Proprietary
Reasoning· Not reasoning
- MAI-Transcribe-2Microsoft · 3 Sep 2026 · — context—in / 1M
- MAI-Cyber-1-FlashMicrosoft · 13 Aug 2026 · — context—in / 1M
- MAI-Image-2.6Microsoft · 10 Aug 2026 · — context—in / 1M
- MAI-Image-2.5-ProMicrosoft · 23 Jul 2026 · — context—in / 1M
- MAI-Voice-2-FlashMicrosoft · 23 Jul 2026 · — context—in / 1M
- MAI-Image-2.5 (and 2.5-Flash)Microsoft · 2 Jun 2026 · — context—in / 1M
- MAI-Transcribe-1.5Microsoft · 2 Jun 2026 · — context—in / 1M
- MAI-Voice-2Microsoft · 2 Jun 2026 · — context—in / 1M
- MAI-Image-2-EfficientMicrosoft · 14 Apr 2026 · — context—in / 1M
- MAI-Transcribe-1Microsoft · 2 Apr 2026 · — context—in / 1M
- MAI-Image-2Microsoft · 19 Mar 2026 · — context—in / 1M
- MAI-Image-1Microsoft · 13 Oct 2025 · — context—in / 1M
- MAI-1-previewMicrosoft · 28 Aug 2025 · — context—in / 1M
- MAI-Voice-1Microsoft · 28 Aug 2025 · — context—in / 1M