Phi-4 — Free AI Model & API

github-models/phi-4
reasoning
Get API key →
Context Window 131K
Max Output 0
Rate Limit See provider page
Cost $0.00 FREE
Free Period Since May 20, 2026
Credit Card Not required
Status Online

Overview

Model ID
Phi-4
Base URL
https://models.inference.ai.azure.com
Specifications
Context: 131K · Output: 0 · Modality: text · OpenAI Compat: Yes

Quick Start

Integrate Phi-4 with 3 lines of code. See the config generator for Claude Code, Cursor, and more.

from openai import OpenAI

client = OpenAI(
 base_url="https://models.inference.ai.azure.com",
 api_key="YOUR_API_KEY"
)

response = client.chat.completions.create(
 model="Phi-4",
 messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)
import OpenAI from "openai";

const openai = new OpenAI({
 baseURL: "https://models.inference.ai.azure.com",
 apiKey: "YOUR_API_KEY",
});

const completion = await openai.chat.completions.create({
 model: "Phi-4",
 messages: [{ role: "user", content: "Hello!" }],
});

console.log(completion.choices[0].message.content);
curl https://models.inference.ai.azure.com/chat/completions \
 -H "Content-Type: application/json" \
 -H "Authorization: Bearer YOUR_API_KEY" \
 -d '{
 "model": "Phi-4",
 "messages": [{"role": "user", "content": "Hello!"}]
 }'

Other Free Models from GitHub Models

gpt-4.1

1.0M context · No card

gpt-4.1-mini

1.0M context · No card

gpt-4o

128K context · No card

o3-mini

200K context · No card

o4-mini

200K context · No card

Rate Limits & Constraints

Rate Limit See provider page
Context Window 131K
Max Output Tokens 0
Cost Free — since May 20, 2026
Credit Card Not required
OpenAI Compatible Yes — drop-in replacement

GitHub Models Platform Limitations

  • Low per-request token limits (8K input / 4K output)
  • Rate limits tied to GitHub Copilot subscription tier
  • Not suitable for large-context or long-generation tasks

Features & Use Cases

Best For

Reasoning

Modality Support

text

GitHub Models Highlights

  • 45+ models including GPT-4.1 and o3
  • Free for all GitHub accounts
  • Includes Llama 4, DeepSeek-R1, Mistral
  • Base URL: models.inference.ai.azure.com

Playground — Test Phi-4

Test Phi-4 directly in your browser. Your API key is sent directly to GitHub Models — never stored.

Model: Phi-4 Get Key

🔒 Your key is never stored — sent directly to the model provider via our server proxy.

Ready to chat with Phi-4.

Frequently Asked Questions

How do I get an API key for Phi-4?

Sign up at GitHub Models to get your API key. No credit card is required — just an email sign-up. Once you have the key, use the code snippets in the Quick Start section above.

Is Phi-4 really free?

Yes. Phi-4 is available on GitHub Models's free tier and has been free since May 20, 2026. Rate limits apply: See provider page. Always check the provider's terms for any changes to the free tier.

What are Phi-4's rate limits?

See provider page Context window: 131K. Max output: 0. No credit card required.

What are the best free alternatives to Phi-4?

Popular free alternatives include NVIDIA: Nemotron 3 Nano Omni (free), Arcee AI: Trinity Large Thinking (free), NVIDIA: Nemotron 3 Super (free). You can also browse all 147+ free models on our site.

More questions? See our full FAQ →

Similar Free Models

NVIDIA: Nemotron 3 Nano Omni (free)

OpenRouter · chat, vision

Arcee AI: Trinity Large Thinking (free)

OpenRouter · chat, reasoning

NVIDIA: Nemotron 3 Super (free)

OpenRouter · chat, reasoning