Home/Models/Phi-3.5 Mini

Phi-3.5 Mini

FreeOpen Source
by Microsoft· Released 2024-08
Official Docs ↗

Phi-3.5 Mini is Microsoft Research's breakthrough small model that achieves remarkable performance for its 3.8B parameter size. It runs on mobile devices and edge hardware, making it ideal for on-device AI applications.

Input Price
$0.10/1M
per 1M tokens
Output Price
$0.10/1M
per 1M tokens
Context Window
128K tokens
max tokens
ELO Score
1112
LMSYS Arena

Pricing Breakdown

VolumeInput CostOutput CostCombined (50/50)
1,000 tokens$0.0001$0.0001$0.0001
10,000 tokens$0.0010$0.0010$0.0010
100,000 tokens$0.0100$0.0100$0.0100
1,000,000 tokens$0.1000$0.1000$0.1000

Strengths

  • Runs on edge devices and smartphones
  • Excellent for its tiny model size
  • 128K context window
  • Free and open source

Weaknesses

  • Significantly weaker than larger models
  • Not suitable for complex reasoning

Best For

fast responselow costcustomer support

Frequently Asked Questions

How much does Phi-3.5 Mini cost?

Phi-3.5 Mini costs $0.10/1M per 1M input tokens and $0.10/1M per 1M output tokens. It is available for free.

What is Phi-3.5 Mini's context window?

Phi-3.5 Mini has a context window of 128K tokens, which means it can process up to 128,000 tokens in a single request.

Is Phi-3.5 Mini open source?

Yes, Phi-3.5 Mini is open source. The model weights are publicly available and can be self-hosted.

What is Phi-3.5 Mini best used for?

Phi-3.5 Mini is best suited for: fast-response, low-cost, customer-support. Runs on edge devices and smartphones.

What is Phi-3.5 Mini's ELO score?

Phi-3.5 Mini has an ELO score of 1112 on the LMSYS Chatbot Arena leaderboard, placing it in the mid-tier range.