Phi 4 API

Microsoft's Phi 4, through an OpenAI-compatible API. Providers quote it on an open exchange, and each request goes to the offer with the lowest estimated charge that meets your rules.

Providers compete for each request. The exchange ranks the offers that meet your rules by estimated charge. It fixes your maximum charge before the first token, and you pay for the tokens that the model uses.
The OpenAI and Anthropic APIs. Use the OpenAI or Anthropic SDK that you have. Set the base URL and your key, and send phi-4 as the model.
Your data rules. Send x-liquid-retention: none and only offers whose provider does not keep your content compete for the request.

Facts

Model name: phi-4
Maker: Microsoft
Context window: 16.4K tokens
Input price: $0.070 per million tokens
Output price: $0.140 per million tokens
Providers: 1 independent provider
Regions: us

Providers

Architect: us, $0.070 input and $0.140 output per million tokens, $0.070 cache read per million tokens, 16.4K tokens of context

Scores

HLE: 3.8, #246 of 273
GPQA: 57.5, #191 of 250
AA-LCR: 0.0, #231 of 237
Terminal-Bench Hard: 3.8, #146 of 166
LiveCodeBench: 23.1, #108 of 130
AIME 2025: 18.0, #87 of 100
MMLU-Pro: 71.4, #86 of 133
τ²-Bench: 0.0, #168 of 169
IFBench: 23.5, #174 of 178

Scores: Artificial Analysis (https://artificialanalysis.ai/)

Request

curl https://router.inference.ai.exchange/v1/chat/completions \
  -H "Authorization: Bearer $LIQUID_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "phi-4",
    "messages": [{"role":"user","content":"Say hello."}]
  }'

Questions

What is Phi 4?: Served by intent-router: each job goes to the cheapest healthy upstream for this model.
How much does Phi 4 cost?: Offers now ask $0.070 per million input tokens and $0.140 per million output tokens. A job pays for the tokens it uses. The exchange picks the cheapest eligible offer.
What is the context length of Phi 4?: The largest offer takes 16,384 input tokens. A smaller offer can take less. The exchange skips an offer that cannot take your message.
What can Phi 4 do?: Its providers list Reasoning effort, JSON object, JSON schema, Frequency penalty, Logit bias, Log probabilities, Min-p sampling, Presence penalty, Repetition penalty, Seed, Stop sequences, Top-k sampling, Top-p sampling, Text, Tools.
What inputs and outputs does Phi 4 support?: Text → Text.
Which providers serve Phi 4?: 1 independent provider quote it now, with 1 offer. The table above ranks them by the cost of a typical job.
How do I call Phi 4?: Send "phi-4" as the model in an OpenAI or Anthropic request. The API section links to setup steps for each client.
When was Phi 4 listed?: The exchange first listed it on 10/8/2026. This is not its release date.