Microsoft Phi API

Add AI to your application with Microsoft Phi.
Access Microsoft Phi through a single API with Puter.js.

Get Started View Tutorial

AI Chat API

Math & Reasoning Powerhouse

Phi-4-reasoning-plus with only 14B parameters outperforms significantly larger models like DeepSeek-R1 70B on complex math and logic tasks.

Unified Multimodal (Vision + Audio + Text)

Phi-4-multimodal handles vision, audio, and text in one model - ideal for speech recognition, translation, and image understanding in a single API call.

Built-in Function Calling

Phi-4 models support function calling to integrate search engines, tools, and external APIs directly.

20+ Languages Supported

Phi models support over 20 languages with strong speech recognition and translation capabilities.

Complex Chat & Agent Workflows

WizardLM-2 8x22B excels at complex chat, multilingual, reasoning, and agent use cases - great for multi-turn conversations and sophisticated assistants.

Cost-Efficient Performance

Small parameter counts (3.8B-14B for Phi) deliver competitive results against much larger models, meaning lower API costs for similar quality.

Microsoft Phi Models

Access all Microsoft Phi models through a single API.

Phi 4 Reasoning Plus

Advanced reasoning model with 14B parameters, excelling at complex math and logic tasks.

microsoft/phi-4-reasoning-plus

Phi 4 Multimodal Instruct

Unified multimodal model handling vision, audio, and text in a single API call.

microsoft/phi-4-multimodal-instruct

Phi 4

Latest generation Phi model with excellent performance for general conversational tasks.

microsoft/phi-4

WizardLM-2 8x22B

Excels at complex chat, multilingual, reasoning, and agent use cases with sophisticated multi-turn conversations.

microsoft/wizardlm-2-8x22b

Add AI Chat in Seconds

Add Microsoft Phi to your app with just a few lines of code.
No API keys, no backend, no configuration required.

<html>
<body>
    <script src="https://js.puter.com/v2/"></script>
    <script>
        puter.ai.chat("Explain quantum computing in simple terms", {
            model: "microsoft/phi-4"
        }).then(response => {
            document.body.innerHTML = response.message.content;
        });
    </script>
</body>
</html>

View full documentation →

Related Resources

Learn how to use Microsoft Phi and other AI models with Puter.js.