LLM EXPLORER 59,420 MODELS INDEXED

Phi 3 Vision 128K Instruct by microsoft

By microsoft · 240595 downloads
HF Repo ↗ Copy URL

Phi 3 Vision 128K Instruct is an open-source language model by microsoft. Features: 4.1b LLM, VRAM: 8.3GB, Context: 128K, License: mit, Instruction-Based, LLM Explorer Score: 0.23.

  Code   Conversational   Custom code   Instruct   Multilingual   Phi3 v   Region:us   Safetensors   Sharded   Tensorflow   Vision
Model Card on HF 🤗: https://huggingface.co/microsoft/Phi-3-vision-128k-instruct 

Phi 3 Vision 128K Instruct Parameters and Internals

Model Type 
multimodal, text generation, vision
Use Cases 
Areas:
commercial, research
Applications:
general image understanding, OCR, chart and table understanding
Primary Use Cases:
memory/compute constrained environments, latency bound scenarios
Limitations:
Not evaluated for all downstream purposes, Potential quality degradation in non-English language use cases
Considerations:
Developers should mitigate biases and misuses.
Supported Languages 
languages_supported (multilingual), proficiency_level (English focused, other languages might perform worse.)
Training Details 
Data Sources:
publicly available documents, high-quality educational data, selected high-quality image-text interleave, synthetic data
Data Volume:
500B vision and text tokens
Methodology:
Supervised fine-tuning, direct preference optimization
Context Length:
128000
Training Time:
1.5 days
Hardware Used:
512 H100-80G GPUs
Model Architecture:
Image encoder, connector, projector, and Phi-3 Mini language model
Safety Evaluation 
Ethical Considerations:
Potential for producing inappropriate, unreliable, or biased content due to training data limitations.
Responsible Ai Considerations 
Fairness:
Potential biases due to varying representation of groups in data.
Transparency:
Developers responsible for evaluating model output for fairness and accuracy.
Accountability:
Developers should ensure model applications adhere to laws and regulations.
Mitigation Strategies:
Safety post-training, adherence to use-case laws.
Input Output 
Input Format:
Single image with chat format prompts
Accepted Modalities:
text, image
Output Format:
Text generated in response to input
LLM NamePhi 3 Vision 128K Instruct
Repository 🤗https://huggingface.co/microsoft/Phi-3-vision-128k-instruct 
Model Size4.1b
Required VRAM8.3 GB
Updated2026-07-31
Maintainermicrosoft
Model Typephi3_v
Instruction-BasedYes
Model Files  4.9 GB: 1-of-2   3.4 GB: 2-of-2
Model ArchitecturePhi3VForCausalLM
Licensemit
Context Length131072
Model Max Length131072
Transformers Version4.38.1
Tokenizer ClassLlamaTokenizer
Padding Token<|endoftext|>
Vocabulary Size32064
Torch Data Typebfloat16

Best Alternatives to Phi 3 Vision 128K Instruct

Best Alternatives
Context / RAM
Downloads
Likes
Phi 3.5 Vision Instruct128K / 8.3 GB1243386737
VLM2Vec Full128K / 8.3 GB57949329
Phi 3.5 Vision Instruct Bf16128K / 8.3 GB482
...on 128K Instruct No Flash Attn128K / 8.3 GB130
Phi 3 Vision 128K Instruct128K / 8.3 GB80
...hi 3 HornyVision 128K Instruct128K / 8.3 GB1729
Phi 3 Vision Win Snap128K / 8.3 GB121
Note: green Score (e.g. "73.2") means that the model is better than microsoft/Phi-3-vision-128k-instruct.