Plain-English Summary
A compact Phi model that can understand text, images, and audio in private or edge deployments.
Phi-4-multimodal-instruct
A compact Phi model that can understand text, images, and audio in private or edge deployments.
Phi-4
Multimodal LLM
GA / open weights
Open weights: Yes
API: Yes and self-hosted
Technical Notes
Multimodal model accepting text, image, and audio with 131K input context and 4K output.
Best For
Document vision, speech and image understanding, private multimodal assistants
Watch Out For
Primarily an understanding model; smaller capacity than large multimodal systems.
Model Specs
- Maker
- Microsoft
- Country
- USA
- Family
- Phi-4
- Type
- Multimodal LLM
- Release date
- 2025-02-26
- Date confidence
- Exact
- Status
- GA / open weights
- Context window
- 131,072 tokens
- Max output
- 4,096
- Parameters
- Not disclosed
- Active parameters
- Not disclosed
- Architecture
- Multimodal transformer
- Reasoning
- General reasoning
- Tool calling
- Host dependent
- Structured output
- Host dependent
- API available
- Yes and self-hosted
- Open weights
- Yes
- Self-hostable
- Yes
- License
- MIT
- Inputs
- Text, image, audio
- Outputs
- Text
- Approx. price
- See provider pricing
- How to access
- Provider website
- Model / API ID
- Phi-4-multimodal-instruct
- Knowledge cutoff
- Not publicly disclosed
- Fine-tuning
- Fine-tuning, quantization, self-hosting
- Completeness
- 90 %
- Verified
- Jul 18, 2026
- Snapshot date
- Jul 18, 2026
- Source
- Official source