Microsoft: AI model benchmarks
AI models from Microsoft. Rank among companies: #107. Strongest model: Phi-3.5-MoE-instruct (Rating 1525, #700 overall). Latest release: MAI-Thinking-1 (Jun 2, 2026).
Models
| Rank | Model | Rating | Released | Access |
|---|---|---|---|---|
| 700 | Phi-3.5-MoE-instruct | 1525 ± 16 | Aug 23, 2024 | Open |
| 703 | WizardLM-2 8x22B | 1523 ± 13 | Apr 15, 2024 | Open |
| 706 | Phi-4 | 1522 ± 5 | Dec 12, 2024 | Open |
| 733 | phi-4-14B | 1516 ± 24 | Jan 16, 2025 | Open |
| 906 | Phi-4 Multimodal Instruct | 1466 ± 6 | Feb 26, 2025 | Open |
| 952 | Phi-3-medium-4k-instruct | 1458 ± 10 | May 21, 2024 | Open |
| 967 | Phi-3-small-8k-instruct | 1456 ± 19 | May 21, 2024 | Open |
| 1013 | Phi-3-medium-128k-instruct | 1448 ± 12 | May 21, 2024 | Open |
| 1105 | Phi-3-small-128k-instruct | 1428 ± 25 | May 21, 2024 | Open |
| 1167 | Phi-3.5-mini-instruct | 1416 ± 9 | Aug 20, 2024 | Open |
| 1236 | WizardLM-70B-V1.0 | 1396 ± 13 | Aug 9, 2023 | Open |
| 1245 | Phi-4 Mini Instruct | 1394 ± 8 | Feb 26, 2025 | Open |
| 1297 | Phi-3 Mini 4K Instruct | 1380 ± 12 | Apr 23, 2024 | Open |
| 1298 | Phi-3 Mini 128K Instruct | 1379 ± 12 | Apr 23, 2024 | Open |
| 1395 | WizardLM-2-7B | 1344 ± 19 | Apr 16, 2024 | Open |
| 1425 | WizardLM-13B-V1.2 | 1332 ± 11 | Jul 25, 2023 | Open |
| 1448 | Orca-2-13B | 1316 ± 34 | Nov 14, 2023 | Open |
| 1486 | WizardLM-13B-V1.0 | 1298 ± 31 | May 13, 2023 | Open |
| 1502 | Orca-2-7B | 1287 ± 16 | Nov 14, 2023 | Open |
| 1511 | Phi-2 | 1279 ± 10 | Dec 12, 2023 | Open |
Interactive version: theaggregate.ai/company?slug=microsoft · How It Works · Data refreshed daily, snapshot 2026-10-20.