Phi-4 Mini Instruct: API pricing, context window and capabilities

Prices from official vendor price lists · October 2026

Phi-4 Mini Instruct costs $0.075 per 1M input tokens and $0.3 per 1M output tokens, with a 131.1K context window. One of the cheapest models here — below 85% of the rating on output price. Suited to high volume, where a wrong answer is cheap to retry.

Visit Microsoft All Microsoft models

Capabilities

API $/1M (in / out)

$0.075 / $0.30

Context window

131.1K tokens

Price / mo

API only

Overview Reviews

Phi-4 Mini Instruct is an AI model from Microsoft, priced at $0.075 input and $0.3 output per 1M tokens. That places it in the budget tier of this rating — cheaper than 85% of the 205 priced models listed.

It holds 131.1K tokens in a single conversation. Capabilities recorded in this rating: text & chat, reasoning, tool calling, low latency, multilingual. Access is through the API rather than a consumer subscription.

API price$0.075 / $0.3per 1M tokens, in / out

Context131.1K tokenstokens per conversation

VendorMicrosoftall models from this vendor

TierBudgetcheaper than 85% of models

Cost calculator

1M tokens/mo10M tokens/mo50M tokens/mo200M tokens/mo

Standard$0.24~~AED 1 per month~~

With batch −50%$0.12~~AED 0 per month~~

With caching on input$0.23~~AED 1 per month~~

Standard$2.44~~AED 9 per month~~

With batch −50%$1.22~~AED 4 per month~~

With caching on input$2.27~~AED 8 per month~~

Standard$12.19~~AED 45 per month~~

With batch −50%$6.09~~AED 22 per month~~

With caching on input$11.34~~AED 42 per month~~

Standard$48.75~~AED 179 per month~~

With batch −50%$24.37~~AED 90 per month~~

With caching on input$45.37~~AED 167 per month~~

Assumes a 1:3 input-to-output ratio, which is typical for chat and generation. Dirham figures use 3.6725 per dollar. Batch and caching discounts apply where the vendor offers them.

Compared with alternatives

How Phi-4 Mini Instruct sits against its neighbours in this rating. Tap a card to open that model.

Hy-MT2-30B-A3B

Cheaper

Tencent

$0.074 / $0.2958.2K tokens

gpt-oss-safeguard-20b

Closest match

OpenAI

$0.075 / $0.3131.1K tokens

Gemma 4 31B

Step up

Google

$0.09 / $0.34262.1K tokens

Good and poor fit

Good fit

Multi-step analysis, planning and code review, where the first plausible answer is often wrong.
High-volume routing, classification and autocomplete.
General text generation and conversation across languages.

Poor fit

Anything requiring image input — this model does not accept it.
Long multi-step reasoning, where a larger model finishes in fewer attempts.

Related filters

Text & chat234 models for text generation and chat

Reasoning175 models that think before answering

Tool calling172 models that call your functions

Low latency76 small, fast models for volume

Open weights94 models you can download and self-host

Microsoft3 models from this vendor

All 245 modelsThe full rating with filters and prices

Frequently asked questions

How much does Phi-4 Mini Instruct cost?

$0.075 / $0.3 per 1M tokens, input and output. In dirhams that is roughly AED 0.28 per 1M input tokens. Batch processing halves it where the vendor offers it.

What is the context window of Phi-4 Mini Instruct?

131.1K tokens in a single conversation.

What can Phi-4 Mini Instruct do?

In this rating it is recorded with: text & chat, reasoning, tool calling, low latency, multilingual, open weights.

Is Phi-4 Mini Instruct a good choice?

It depends on the job. See the good-fit and poor-fit lists above, and compare against the neighbouring models before committing.

Ratings 0

Avg. score 0

Rate this page

Help us improve