Skip to content
RankShark

Phi 4

[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with

ModelOpen sourceby Microsoftsince 2025
38.7

SharkScore

#108 in Models

Why it scores 38.7

Phi 4 has a SharkScore of 38.7/100, ranked #108 of 606 in Models (updated 2026-11-18).

Each part compares Phi 4 with other listings in Models. Evidence covers 87% of the score's weight.

Adoption
22 · 30%
User reviews
no data · 25%
Maintenance
27 · 20%
Trust & transparency
80 · 15%
Momentum
no data · 10%

Paying never changes this score. How ranking works · method v2026-11-18.2

RankShark review of Phi 4

Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review

We were not paid for this review.

As of 2026-11-18, Phi 4 is Microsoft's 14B text model; OpenRouter lists a 16,384-token context and measured rates of $0.07/M input and $0.14/M output. [E1, E2, MEASURED]

Phi-4 is Microsoft's 14B text-generation model, released under the MIT license. Its model card describes it as primarily English-focused and best suited to chat-format prompts. [E1]

Best for

  • Developers building English-language generative AI features where memory or compute is constrained. · model card
  • Developers exploring reasoning, math, or code-generation tasks with a text model. · model card

Not for

  • Teams seeking a model intended for multilingual use; the model card says Phi-4 is not intended to support multilingual use. · model card
  • Applications that require consistently accurate outputs without additional evaluation; the model card warns that outputs can be inaccurate or outdated. · model card

Key features

  • Model architecture. The vendor says Phi-4 has 14B parameters and uses a dense decoder-only Transformer architecture. · model card
  • Intended use. The vendor says the model is intended for general-purpose AI systems, including memory- or compute-constrained and latency-bound scenarios, and reasoning tasks. · model card
  • Context window. The OpenRouter listing has a measured context window of 16,384 tokens. · OpenRouter
  • License. The vendor lists the model under the MIT license. · model card

Pros

  • The vendor lists the model under the MIT license. · model card
  • The OpenRouter listing has measured rates of $0.07 per million input tokens and $0.14 per million output tokens. · OpenRouter
  • The model card provides example usage with Transformers. · model card

Cons

  • The vendor says the model is not intended for multilingual use and that languages other than English will perform worse. · model card
  • The vendor warns that outputs can be inaccurate or outdated. · model card
  • The vendor says most code-training data is based on Python and recommends manually verifying API use in generated scripts. · model card
  • No free plan or free trial stated on the pages we checked.
  • Whether customer data is used for model training is not stated on the pages we checked.

Pricing

PlanPriceLimits
Not stated$0.07 per million input tokens; $0.14 per million output tokens on the OpenRouter listing.16,384-token context window (measured). No plan limits stated.

Checked 18 Nov 2026

Free plan
Not stated (checked 18 Nov 2026)
Free trial
Not stated (checked 18 Nov 2026)
Trains on customer data
Not stated (checked 18 Nov 2026)
Data retention
Not stated (checked 18 Nov 2026)
Data regions
Not stated (checked 18 Nov 2026)
Data processing agreement
Not stated (checked 18 Nov 2026)
AI models used
Phi-4, a 14B-parameter model. · model card
API
The OpenRouter listing provides chat completions, completions, messages, and responses endpoints. · OpenRouter
SDKs
Not stated (checked 18 Nov 2026)
Webhooks
Not stated (checked 18 Nov 2026)
Platforms
Hugging Face model repository; the model card includes Transformers usage. · model card
Security claims
No security certification stated (checked 18 Nov 2026)

What we measured

Context window (OpenRouter listing) · now
16,384 tokens
Price per million tokens (OpenRouter listing) · now
$0.07 in, $0.14 out
Package downloads · 30 days
205,674
How we tested · sources
  • Hourly uptime and response-time checks of the product's URL.
  • Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
  • We did not sign up, pay, or use the product as a customer.
  • 18 Nov 2026: Plans or prices changed on the pricing page.
  • 18 Nov 2026: Plans or prices changed on the pricing page.

Something wrong? Report an error with a public link that shows it; corrections re-run the review.

Top alternatives to Phi 4

See all
  • 67.1

    Qwen3.8 27B is an open-weight dense vision-language model from Qwen.

    ModelOpen source
  • 65.5

    Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per t

    ModelOpen source
  • 64.7

    Qwen3.8 Flash is a multimodal reasoning model from Alibaba.

    ModelOpen source
  • 64.3

    Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026.

    ModelOpen source