Identity
- Type
- big tech
- Country label
- USA
- HQ country
- USA
- Founding country
- USA
- Parent
- not in the notes
- Parent country
- not in the notes
- Country basis
- the research notes
- Units
GitHub
Where this comes from
Convention D1: a dual label, HQ first, when the HQ, founding or parent country differ.
Models by Class
Models
ModelClassOpennessParametersReleased
MAI-Transcribe-2speech recognition3 Sep 2026
MAI-Image-2.6image generationAug 2026
LightGBMtabular18 Jul 2026
GridSFM Opengraph12 May 2026
MAI-Thinking-1reasoningMay 2026
MAI-Transcribe-1speech recognition2 Apr 2026
harrier-oss-v1-0.6bembedding30 Mar 2026
Phi-4-reasoning-vision-15Bmultimodal LLM4 Mar 2026
VibeVoice-ASRspeech recognition21 Jan 2026
TRELLIS.2-4B3D16 Dec 2025
Phi-4-miniedge LLMDec 2025
Fara-7Bmultimodal LLM30 Oct 2025
MAI-Voice-1speech synthesis28 Aug 2025
Dayhoff-3b-GR-HM-cscientific4 Jul 2025
Phi-4-mini-flash-reasoningreasoning19 Jun 2025
Auroraweather21 May 2025
Magma-8Bmultimodal LLM23 Feb 2025
Phi-4-mini-instructedge LLM19 Feb 2025
BioEmuscientific19 Feb 2025
Phi-4-multimodal-instructmultimodal LLM2025
TRELLIS (image-large)3D2 Dec 2024
LLM2CLIPvision7 Nov 2024
MedImageParse (BiomedParse)vision4 Nov 2024
Phi-3.5-vision-instructmultimodal LLM16 Aug 2024
Florence-2 basemultimodal LLM15 Jun 2024
Florence-2 largemultimodal LLM15 Jun 2024
LLaVA-NeXT-Video 7BUniversity of Wisconsin-Madison, Microsoft AImultimodal LLM5 Jun 2024
MAIRA-2multimodal LLMJun 2024
RAD-DINOvision17 May 2024
LLaVA-Med v1.5 (Mistral 7B)multimodal LLM14 May 2024
MatterSimmaterials8 May 2024
Phi-3LLMApr 2024
Phi-3-miniedge LLMApr 2024
LLMLingua-2language encoder17 Mar 2024
LLaVA-NeXT (1.6) Mistral 7BUniversity of Wisconsin-Madison, Microsoft AImultimodal LLM20 Feb 2024
e5-mistral-7bembedding2024
multilingual-e5-large-instructembedding2024
Phi-2edge LLM13 Dec 2023
MatterGenmaterials6 Dec 2023
LLaVA-1.5 7BUniversity of Wisconsin-Madison, Microsoft AImultimodal LLM5 Dec 2023
Florence-2multimodal LLM10 Nov 2023
Kosmos-2multimodal LLM2 Oct 2023
multilingual-e5-smallembedding30 Jun 2023
LLaVAUniversity of Wisconsin-Madison, Microsoft AImultimodal LLM17 Apr 2023
BiomedCLIPembedding2 Mar 2023
VALL-Espeech synthesis5 Jan 2023
Graphormer base (PCQM4Mv2)graph5 Jan 2023
BioGPTLLM20 Nov 2022
BEiT-3vision22 Aug 2022
BEiT v2vision12 Aug 2022
X-CLIPvision4 Aug 2022
X-CLIP base patch32embedding4 Aug 2022
LayoutLMv3OCR and document18 Apr 2022
DiT (Document Image Transformer)OCR and document4 Mar 2022
TAPEX base fine-tuned on WikiSQLtabular2 Mar 2022
Swin Transformer V2vision18 Nov 2021
DeBERTa-v3 baselanguage encoder18 Nov 2021
DeBERTa-v3 largelanguage encoder18 Nov 2021
WavLMspeech recognition26 Oct 2021
wavlm-base-plusspeech recognition26 Oct 2021
wavlm-largespeech recognition26 Oct 2021
SpeechT5speech synthesis14 Oct 2021
Table TransformerOCR and document30 Sep 2021
TrOCROCR and document21 Sep 2021
TrOCR small handwrittenOCR and document21 Sep 2021
TrOCR large printedOCR and document21 Sep 2021
Conditional DETRvision13 Aug 2021
TAPEXtabular16 Jul 2021
BEiTvision15 Jun 2021
BEiT-Bvision15 Jun 2021
Graphormergraph9 Jun 2021
LayoutXLMOCR and document18 Apr 2021
CvTvision29 Mar 2021
Swin Transformervision25 Mar 2021
Swin-Bvision25 Mar 2021
Swin-Tvision25 Mar 2021
LayoutLMv2OCR and document29 Dec 2020
BiomedBERT (PubMedBERT)language encoder31 Jul 2020
FastSpeech 2speech synthesis8 Jun 2020
DeBERTalanguage encoder5 Jun 2020
MPNet baselanguage encoder20 Apr 2020
MiniLMlanguage encoder25 Feb 2020
LayoutLMOCR and document31 Dec 2019
Spectral Residual (SR and SR-CNN)time series10 Jun 2019
MAI-Imageimage generation–
MAI-Image-2.5image generation–
Azure AI Content Safetyguard–
MAI-Code-1.1-Flashcode–
MAI-Cyber-1-FlashLLM–
MAI-Voice-2.1speech synthesis–
mDeBERTa-v3-baseembedding–
Phiedge LLM–
Phi-3.5-miniLLM–
phi-4LLM–
Phi-4-mini-reasoningreasoning–
Phi-4-reasoningreasoning–
Prov-GigaPathMicrosoft AI, Providencevision–
VibeVoice-1.5Bspeech synthesis–
Siemens Industrial Foundation Model (IFM)Siemens, Microsoft AImultimodal LLM–
Bringing Old Photos Back to Lifevision–
Explainable Boosting Machine (EBM)tabular–
EvoDiffscientific–Links
Organisation
Microsoft AIbig tech, USA
Microsoft AIbig tech, USAModels
MAI-Transcribe-2Microsoft AI, speech recognition
MAI-Image-2.6Microsoft AI, image generation
LightGBMMicrosoft AI, tabular
GridSFM OpenMicrosoft AI, graph+98 more
MAI-Transcribe-2Microsoft AI, speech recognition
MAI-Image-2.6Microsoft AI, image generation
LightGBMMicrosoft AI, tabular
GridSFM OpenMicrosoft AI, graph+98 moreModels 102
LayoutLMMicrosoft AI, OCR and documentDeveloper
LLaVAUniversity of Wisconsin-Madison, Microsoft AI, multimodal LLMDeveloper
Florence-2Microsoft AI, multimodal LLMDeveloper
phi-4Microsoft AI, LLMDeveloper
BEiTMicrosoft AI, visionDeveloper
BEiT v2Microsoft AI, visionDeveloper
BEiT-3Microsoft AI, visionDeveloper
Florence-2 baseMicrosoft AI, multimodal LLMDeveloper
Florence-2 largeMicrosoft AI, multimodal LLMDeveloper
LLaVA-NeXT-Video 7BUniversity of Wisconsin-Madison, Microsoft AI, multimodal LLMDeveloper
MAIRA-2Microsoft AI, multimodal LLMDeveloper
PhiMicrosoft AI, edge LLMDeveloper+90
Phi-3Microsoft AI, LLMDeveloper
Phi-3.5-vision-instructMicrosoft AI, multimodal LLMDeveloper
Phi-4-mini-instructMicrosoft AI, edge LLMDeveloper
Swin TransformerMicrosoft AI, visionDeveloper
BEiT-BMicrosoft AI, visionDeveloper
DeBERTaMicrosoft AI, language encoderDeveloper
e5-mistral-7bMicrosoft AI, embeddingDeveloper
Kosmos-2Microsoft AI, multimodal LLMDeveloper
LayoutLMv2Microsoft AI, OCR and documentDeveloper
LayoutLMv3Microsoft AI, OCR and documentDeveloper
LLaVA-1.5 7BUniversity of Wisconsin-Madison, Microsoft AI, multimodal LLMDeveloper
LLaVA-Med v1.5 (Mistral 7B)Microsoft AI, multimodal LLMDeveloper
LLaVA-NeXT (1.6) Mistral 7BUniversity of Wisconsin-Madison, Microsoft AI, multimodal LLMDeveloper
MAI-ImageMicrosoft AI, image generationDeveloper
MAI-Image-2.5Microsoft AI, image generationDeveloper
MAI-Image-2.6Microsoft AI, image generationDeveloper
multilingual-e5-smallMicrosoft AI, embeddingDeveloper
Phi-3-miniMicrosoft AI, edge LLMDeveloper
Phi-3.5-miniMicrosoft AI, LLMDeveloper
Phi-4-mini-flash-reasoningMicrosoft AI, reasoningDeveloper
Phi-4-mini-reasoningMicrosoft AI, reasoningDeveloper
Phi-4-reasoningMicrosoft AI, reasoningDeveloper
Prov-GigaPathMicrosoft AI, Providence, visionDeveloper
Siemens Industrial Foundation Model (IFM)Siemens, Microsoft AI, multimodal LLMDeveloper
Swin Transformer V2Microsoft AI, visionDeveloper
Swin-BMicrosoft AI, visionDeveloper
TrOCRMicrosoft AI, OCR and documentDeveloper
WavLMMicrosoft AI, speech recognitionDeveloper
BiomedCLIPMicrosoft AI, embeddingDeveloper
DeBERTa-v3 baseMicrosoft AI, language encoderDeveloper
DeBERTa-v3 largeMicrosoft AI, language encoderDeveloper
DiT (Document Image Transformer)Microsoft AI, OCR and documentDeveloper
Fara-7BMicrosoft AI, multimodal LLMDeveloper
GraphormerMicrosoft AI, graphDeveloper
Graphormer base (PCQM4Mv2)Microsoft AI, graphDeveloper
harrier-oss-v1-0.6bMicrosoft AI, embeddingDeveloper
LayoutXLMMicrosoft AI, OCR and documentDeveloper
Magma-8BMicrosoft AI, multimodal LLMDeveloper
MAI-Code-1.1-FlashMicrosoft AI, codeDeveloper
MatterGenMicrosoft AI, materialsDeveloper
mDeBERTa-v3-baseMicrosoft AI, embeddingDeveloper
MedImageParse (BiomedParse)Microsoft AI, visionDeveloper
multilingual-e5-large-instructMicrosoft AI, embeddingDeveloper
Phi-4-multimodal-instructMicrosoft AI, multimodal LLMDeveloper
Phi-4-reasoning-vision-15BMicrosoft AI, multimodal LLMDeveloper
RAD-DINOMicrosoft AI, visionDeveloper
Swin-TMicrosoft AI, visionDeveloper
Table TransformerMicrosoft AI, OCR and documentDeveloper
TAPEXMicrosoft AI, tabularDeveloper
TAPEX base fine-tuned on WikiSQLMicrosoft AI, tabularDeveloper
TRELLIS (image-large)Microsoft AI, 3DDeveloper
TRELLIS.2-4BMicrosoft AI, 3DDeveloper
TrOCR large printedMicrosoft AI, OCR and documentDeveloper
TrOCR small handwrittenMicrosoft AI, OCR and documentDeveloper
wavlm-base-plusMicrosoft AI, speech recognitionDeveloper
wavlm-largeMicrosoft AI, speech recognitionDeveloper
X-CLIPMicrosoft AI, visionDeveloper
X-CLIP base patch32Microsoft AI, embeddingDeveloper
AuroraMicrosoft AI, weatherDeveloper
Azure AI Content SafetyMicrosoft AI, guardDeveloper
BioEmuMicrosoft AI, scientificDeveloper
BioGPTMicrosoft AI, LLMDeveloper
BiomedBERT (PubMedBERT)Microsoft AI, language encoderDeveloper
Bringing Old Photos Back to LifeMicrosoft AI, visionDeveloper
Conditional DETRMicrosoft AI, visionDeveloper
CvTMicrosoft AI, visionDeveloper
Dayhoff-3b-GR-HM-cMicrosoft AI, scientificDeveloper
EvoDiffMicrosoft AI, scientificDeveloper
Explainable Boosting Machine (EBM)Microsoft AI, tabularDeveloper
FastSpeech 2Microsoft AI, speech synthesisDeveloper
GridSFM OpenMicrosoft AI, graphDeveloper
LightGBMMicrosoft AI, tabularDeveloper
LLM2CLIPMicrosoft AI, visionDeveloper
LLMLingua-2Microsoft AI, language encoderDeveloper
MAI-Cyber-1-FlashMicrosoft AI, LLMDeveloper
MAI-Thinking-1Microsoft AI, reasoningDeveloper
MAI-Transcribe-1Microsoft AI, speech recognitionDeveloper
MAI-Transcribe-2Microsoft AI, speech recognitionDeveloper
MAI-Voice-1Microsoft AI, speech synthesisDeveloper
MAI-Voice-2.1Microsoft AI, speech synthesisDeveloper
MatterSimMicrosoft AI, materialsDeveloper
MiniLMMicrosoft AI, language encoderDeveloper
MPNet baseMicrosoft AI, language encoderDeveloper
Phi-2Microsoft AI, edge LLMDeveloper
Phi-4-miniMicrosoft AI, edge LLMDeveloper
Spectral Residual (SR and SR-CNN)Microsoft AI, time seriesDeveloper
SpeechT5Microsoft AI, speech synthesisDeveloper
VALL-EMicrosoft AI, speech synthesisDeveloper
VibeVoice-1.5BMicrosoft AI, speech synthesisDeveloper
VibeVoice-ASRMicrosoft AI, speech recognitionDeveloperOrganisations 1
Scores
| Benchmark | Model | Score | Who ran it | Setting |
|---|---|---|---|---|
| AI2D | Phi-4-reasoning-vision-15B | 84.8 % | vendor-reported | |
| AIME 2025 | Phi-4-mini-flash-reasoning | 33.59 % | vendor-reported | |
| AIME24 | Phi-4-mini-flash-reasoning | 52.29 % | vendor-reported | |
| Artificial Analysis Text-to-Image Arena | MAI-Image-2.6 | 1,151 Elo | independent | rank shown as 5-6 |
| GPQA | Phi-4-mini-instruct | 25.2 % | vendor-reported | 0-shot CoT |
| GPQA Diamond | Phi-4-mini-flash-reasoning | 45.08 % | vendor-reported | |
| GSM8K | Phi-4-mini-instruct | 88.6 % | vendor-reported | |
| Hugging Face Open ASR Leaderboard | Phi-4-multimodal-instruct | 6.02 WER | maintainer-run | eight-set macro-average English WER |
| LMArena Text-to-Image Arena | MAI-Image-2.6 | 1,332 Elo | independent | |
| MATH | Phi-4-mini-instruct | 64 % | vendor-reported | |
| MATH500 | Phi-4-mini-flash-reasoning | 92.45 % | vendor-reported | |
| MMLU | Phi-3-mini | 69 % | vendor-reported | |
| MMLU | Phi-4-mini-instruct | 67.3 % | vendor-reported | 5-shot |
| MMLU-Pro | Phi-4-mini-instruct | 52.8 % | vendor-reported | |
| MMMU | Phi-4-reasoning-vision-15B | 54.3 % | vendor-reported | |
| Multilingual MMLU | Phi-4-mini-instruct | 49.3 % | vendor-reported | |
| TabArena | LightGBM | 1,188 Elo | maintainer-run | default |
| TabArena | LightGBM | 1,387 Elo | maintainer-run | tuned |
| TabArena | LightGBM | 1,435 Elo | maintainer-run | tuned + ensembled |
| WeatherBench 2 | Aurora | 1.501 RMSE K | maintainer-run | T850 day 5, deterministic, Aurora (oper.) vs Analysis |
| WeatherBench 2 | Aurora | 119.8 RMSE m2/s2 | maintainer-run | Z500 day 3, deterministic, Aurora (oper.) vs Analysis |
| WeatherBench 2 | Aurora | 2.979 RMSE K | maintainer-run | T850 day 10, deterministic, Aurora (oper.) vs Analysis |
| WeatherBench 2 | Aurora | 262.8 RMSE m2/s2 | maintainer-run | Z500 day 5, deterministic, Aurora (oper.) vs Analysis |
| WeatherBench 2 | Aurora | 38.1 RMSE m2/s2 | maintainer-run | Z500 day 1, deterministic, Aurora (oper.) vs Analysis |
| WeatherBench 2 | Aurora | 668.4 RMSE m2/s2 | maintainer-run | Z500 day 10, deterministic, Aurora (oper.) vs Analysis |
Where this comes from
Every recorded score of this organisation's models, with who ran it. Scores from different settings are not directly comparable.
Sources
- Microsoft AI: News microsoft.ai https://microsoft.ai/news/
- Models | Microsoft AI microsoft.ai https://microsoft.ai/models/
- huggingface.co/microsoft/Phi-4-reasoning-vision-15B https://huggingface.co/microsoft/Phi-4-reasoning-vision-15B
- What is Azure AI Content Safety? - Azure AI services learn.microsoft.com https://learn.microsoft.com/en-us/azure/ai-services/content-safety/overview
- Releases , lightgbm-org/LightGBM github.com https://github.com/microsoft/LightGBM/releases
- Aurora - Microsoft Foundry Labs labs.ai.azure.com https://labs.ai.azure.com/innovations/aurora/
- Microsoft - Wikipedia en.wikipedia.org https://en.wikipedia.org/wiki/Microsoft
- microsoft / phi-4-mini-flash-reasoning docs.api.nvidia.com https://docs.api.nvidia.com/nim/reference/microsoft-phi-4-mini-flash-reasoning
- GitHub - microsoft/onnxruntime: ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator github.com https://github.com/microsoft/onnxruntime
- huggingface.co/microsoft/maira-2 https://huggingface.co/microsoft/maira-2
- MAIRA-2: Grounded Radiology Report Generation - Microsoft Research microsoft.com https://www.microsoft.com/en-us/research/publication/maira-2-grounded-radiology-report-generation/
- Microsoft introduces new adapted AI models for industry - The Official Microsoft Blog blogs.microsoft.com https://blogs.microsoft.com/blog/2024/11/13/microsoft-introduces-new-adapted-ai-models-for-industry/
- NSF launches a new AI research operations center nextgov.com https://www.nextgov.com/artificial-intelligence/2026/09/nsf-launches-new-ai-research-operations-center/415786/
- Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone arxiv.org https://arxiv.org/abs/2404.14219
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Microsoft AI", https://aurion.semai.id/organisations/microsoft/, records of October 2026.


