Cracked AI Engineering

Phi-3.5-vision instruct (128k)

GitHub ModelsOpen Weights

The Phi-3.5-vision instruct (128k) AI model by GitHub Models offers developers a powerful tool with a context window of 128,000 tokens and the ability to process text, vision, function calling, and advanced reasoning tasks. With open weights and no cost for input or output tokens, this model is ideal for a wide range of applications. Take advantage of its cutting-edge capabilities for your projects before the training data cutoff in October 2023.

Key Specifications

Context Window

128K tokens

Max Output Tokens

4K tokens

Input Pricing

Free

per million tokens

Output Pricing

Free

per million tokens

Capabilities

  • Text Generation
  • Vision
  • Function Calling
  • Advanced Reasoning

Additional Details

Release Date
August 20, 2024
Advanced Reasoning
Supported
Supported Input Types
text, image

Compare with Similar Models