Vendor · Language models · Image models · Audio models · Specialist models
Microsoft has 21 models in the directory: 8 language models, 5 image models, 7 audio models and 1 specialist model.
Azure, GitHub Copilot, and Microsoft 365 Copilot. At the close of OpenAI’s recapitalization on October 28, 2025, Microsoft held roughly 27% of OpenAI Group, according to OpenAI. This is a dated snapshot of that structure, not a current ownership breakdown.
In the directory: 8 language models, 5 image models, 7 audio models and 1 specialist model. 6 of them with open weights that can run locally.
Every model there ever was. Models with a profile lead to every benchmark, price and source.
| MAI-Thinking-1 | Aug 12, 2026 | Preview | n/a | n/a | n/a | 256K | below 1 |
| MAI-Code-1.1-Flash | Aug 11, 2026 | Active | n/a | n/a | n/a | n/a | 1 |
| Phi-4-mini | Feb 26, 2025 | Active | n/a | n/a | n/a | 128K | 42 |
| Phi-4-multimodal-instruct | Feb 26, 2025 | Active | n/a | n/a | n/a | 131K | 54 |
| Phi-4 | December 2024 | Active | n/a | $0.056 / $0.11 | 42 | 16K | 57 |
| Phi-3 Medium (14B) | May 2024 | Active | n/a | n/a | n/a | 131K | 30 |
| Phi-3 minino profile | April 2024 | No status data | n/a | n/a | n/a | n/a | n/a |
| WizardLM-2 8x22B | April 2024 | Active | n/a | n/a | n/a | 66K | 37 |
| MAI-Image-2.6 | Jul 31, 2026 | Preview | $0.039 per image | 45 |
| MAI-Image-2.6 Flash | Jul 31, 2026 | Preview | $0.019 per image | below 1 |
| MAI-Image-2.5-Pro | Jul 23, 2026 | Preview | n/a | below 1 |
| MAI-Image-2.5 | June 2026 | Preview | $0.048 per image | 43 |
| MAI-Image-2.5 Flash | June 2026 | Preview | $0.02 per image | 33 |
| MAI-Transcribe-2-Streaming | Oct 1, 2026 | Preview | n/a | 3 |
| MAI-Voice-2.1 | Oct 1, 2026 | Preview | n/a | 2 |
| MAI-Voice-2.1-Flash | Oct 1, 2026 | Preview | n/a | 1 |
| MAI-Transcribe-2 | Sep 3, 2026 | Preview | n/a | below 1 |
| MAI-Voice-2-Flash | Jul 23, 2026 | Preview | n/a | below 1 |
| MAI-Transcribe-1.5 | Jun 2, 2026 | Preview | n/a | below 1 |
| MAI-Voice-2 | Jun 2, 2026 | Active | n/a | below 1 |
| MAI-Cyber-1-Flash | Aug 13, 2026 | Limited access | n/a | below 1 |
Language model · Aug 11, 2026
MAI-Code-1.1-Flash reduces token use on coding tasks. Microsoft reports 25% fewer tokens per task and 25% faster token streaming in GitHub Copilot compared with version 1.0. Its post-training focuses in part on CLI tasks and .NET.
Audio model · Jun 2, 2026
MAI-Voice-2 is Microsoft’s earlier Foundry speech synthesis model with 15 supported languages. A 5 to 60 second reference recording can guide voice cloning, with Microsoft describing consent guardrails.
Language model · Feb 26, 2025
New in Phi-4-mini are a vocabulary enlarged to about 200,000 tokens, grouped-query attention, and shared input and output embeddings. Microsoft lists 23 supported languages. The open 3.8B instruction checkpoint has a 128K-token context and was trained with synthetic and filtered public data, SFT, and DPO.
Language model · Feb 26, 2025
Phi-4-multimodal-instruct processes text, images and audio and answers in text. The model has 5.6 billion parameters and uses Phi-4-mini as its language backbone, extended with encoders and adapters for vision and speech.
Language model · December 2024
Phi-4 combines 14 billion dense parameters with a 16,384-token context. Microsoft trained the checkpoint on 9.8 trillion tokens, including a substantial share of synthetic and curated data. The weights use the MIT license.
Language model · May 2024
Phi-3 Medium (14B) is Microsoft’s open instruction model for substantial capability under constrained compute and latency budgets. The model card distinguishes 4K and 128K variants. The long-context 128K checkpoint uses synthetic and filtered public data followed by SFT and DPO, described as targeting code, mathematics, logic, and instruction following.
Language model · April 2024
WizardLM-2 8x22B is Microsoft’s large sparse Mixture-of-Experts instruction model. The checkpoint has 141 billion total parameters, with roughly 39 billion activated per token. Microsoft positions it as the largest WizardLM-2 variant for complex conversation, multilingual work, reasoning, and agentic tasks. Its context window is 65,536 tokens.
| Vendor | Models | Model types | Best language model | Top 10 | Open weights | New in 2026 |
|---|---|---|---|---|---|---|
| Anthropic | 33 | 33 language models | Claude Opus 5.5 (Rank 1 to 3) | 4 | 0 | 13 |
| OpenAI | 63 | 46 language models, 4 image models, 2 video models, 9 audio models, 2 specialist models | GPT-6 Astra (Rank 4 to 8) | 2 | 4 | 27 |
| 66 | 43 language models, 4 image models, 6 video models, 11 audio models, 2 specialist models | Gemini 4 Carbon (Rank 1 to 5) | 2 | 19 | 26 | |
| Meta | 32 | 29 language models, 1 image model, 1 video model, 1 specialist model | Muse Wonder Full (Rank 6 to 9) | 2 | 26 | 7 |
| DeepSeek | 23 | 23 language models | DeepSeek V4.1 Flash 1017 (Rank 13 to 20) | 0 | 23 | 9 |
| Moonshot AI | 12 | 12 language models | Kimi K3.1 Pro (Rank 10 to 13) | 0 | 10 | 7 |
| xAI | 16 | 12 language models, 2 image models, 2 video models | Grok 4.7 (Rank 18 to 24) | 0 | 1 | 12 |
| Microsoft | 21 | 8 language models, 5 image models, 7 audio models, 1 specialist model | not ranked yet | 0 | 6 | 15 |
By release date. Grey: retired models.
15 releases since January 2026, by release date.
Prices per 1M tokens for the current language models, according to Microsoft or the provider linked in each model profile.
| Current model | Input | Output | Context |
|---|---|---|---|
| Phi-4 | $0.056 | $0.11 | 16K |
| Current model | Price |
|---|---|
| MAI-Image-2.6 | $0.039 per image |
| MAI-Image-2.6 Flash | $0.019 per image |
| MAI-Image-2.5 | $0.048 per image |
| MAI-Image-2.5 Flash | $0.02 per image |
Monthly prices in Germany incl. VAT; “≈” is converted from US dollars. The vendor's models are often also part of other providers' plans, see the AI subscription comparison.
| Plan | Per month | Note |
|---|---|---|
| Microsoft 365 Premium | €22 | |
| Microsoft 365 Pro | €110 |
The directory lists 21 Microsoft models: 8 language models, 5 image models, 7 audio models and 1 specialist model. It counts every model Microsoft ever released, including retired ones; 20 of them have their own profile. 7 are currently available, the newest being MAI-Code-1.1-Flash, MAI-Voice-2 and Phi-4-mini.
MAI-Transcribe-2-Streaming, released on Oct 1, 2026. The timeline below shows every release by date.
Through the API per 1M tokens (input / output): Phi-4 $0.056 / $0.11. The prices of all 1 current language models are in the price table. As a subscription, Microsoft models start at €22 per month incl. VAT (Microsoft 365 Premium).
The AI subscription comparison lists no free plan for Microsoft. The cheapest plan is Microsoft 365 Premium at €22 per month incl. VAT.
Yes, 6 of the 21 models have open weights. Each model profile lists how much memory it needs.
Phi-4 with 42 tokens per second according to Artificial Analysis. A typical model manages 98.
The largest is MAI-Thinking-1 with 256,000 tokens (≈ 380 A4 pages). Every model's value is in the model list.
Data as of Nov 1, 2026