Phi-4 Reasoning — Microsoft | Modeldex

Phi-4 Reasoning

Active

14B reasoning-specialized Phi model with extended thinking.

Updated 3 days agoStructured data from Modeldex catalog

Open source

Not enough benchmark coverage yet for an Intelligence Index — needs at least 3 results across 2 categories.

Overview

Phi-4 Reasoning applies reinforcement learning on top of the Phi-4 base to produce a small model with chain-of-thought capabilities competitive with much larger models on math and science tasks.

History

Phi-4 Reasoning was released on 2025-04-01.

Training & availability

Weights are publicly available under the MIT license, making this an open-weight model suitable for on-prem deployment and fine-tuning.

Capabilities

Context window: 32K tokens.
Max output: 16K tokens.
Input modalities: text.

Recommended for: open-source.

Limitations

The context window (32K tokens) is modest by 2026 standards — unsuitable for processing long documents in a single request.
Text-only — cannot process images, audio, or video inputs.

Quick start

Minimal example using the OpenRouter API. Copy, paste, replace the key.

from openai import OpenAI

client = OpenAI(
    base_url="https://openrouter.ai/api/v1",
    api_key="sk-or-...",
)
resp = client.chat.completions.create(
    model="microsoft/phi-4-reasoning",
    messages=[{"role": "user", "content": "Explain quantum computing in one sentence."}],
)
print(resp.choices[0].message.content)

Cost calculator

Estimate your monthly bill. Presets are typical workload sizes.

Input tokens / month5.0M

@ $0.075/1M

Output tokens / month2.0M

@ $0.3/1M

Input cost

$0.375

5.0M × $0.075/1M

Output cost

$0.6

2.0M × $0.3/1M

Total / month

$0.975

$11.7 / year

Integrations & tooling support

Tool calling: Not supported
Structured outputs: Not supported

Price vs quality

Budget pricing

Priced low — good for high-volume tasks. Quality tier pending more benchmark coverage.