Microsoft: Phi 4

microsoft/phi-4

[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion...

Modalities

Text → Text

Input / Output Price$0.07 / $0.14From · per 1M tokens · see provider prices

Context16.4Ktokens · maximum across providers

ReleasedJan 10, 2025Public catalog metadata

Availability & data sources Listed in CLSSAI · 1 reference endpoints

CLSSAI API · Directory listedCatalog preview · Exact metadata

ACCESS

Where this model works

Catalog preview and CLSSAI API routing are independent decisions with separate sources.

Catalog previewTestable in Playground

Testable

Exact capability metadata is available. CLSSAI can prepare the Text chat request without guessing the model contract.

Evidence: Exact ID microsoft/phi-4 · /chat/completions
Mapped mode: Text chat

CLSSAI APICallable through CLSSAI API

Callable

The exact model ID is listed in the current CLSSAI public directory. This proves current route eligibility, not a completed inference for every mode.

Evidence: Exact ID microsoft/phi-4 is directory-listed

Text test status

The request contract and frontend path are ready. A real recorded output has not yet been captured for this exact model and mode.

Request contract: Build verified
/chat/completions · Schema tests · 2026-08-05
Frontend path: Available
Text form · 2026-08-05
Real model output: Not yet verified
No keyed smoke-test record

Providers

Compare published prices and endpoint capabilities. Select a provider for context limits, caching prices and supported parameters.


	$0.07	$0.14	--	--	--	100.00%

USD per 1M tokens. Prices and metrics describe catalog endpoints, not CLSSAI routing. “--” means no measurement was supplied. Uptime: last 30 minutes.

Pricing

Published base rates across providers, in USD per 1M tokens. Actual cost depends on the provider, caching and prompt length.

Input / 1M tokens$0.07

Output / 1M tokens$0.14

Cache read / 1M tokens--

Effective Pricing: request-weighted costs and historical prices are not supplied by the public source.

Performance

Latency and throughput history are not supplied by the public endpoint API. Check the catalog for its latest performance charts.

Uptime

Successful requests in the last 30 minutes, as reported by the catalog. A snapshot does not show historical reliability.

DeepInfra Bf16100.00%

Benchmarks

No scored evaluations are included in the public model metadata. View the source for benchmarks and their methodology.

Apps

Application rankings require verified traffic data, which is not included in this source.

Activity

Historical token and request volumes are not supplied by the public API. The catalog shows activity on its own network.

Frequently asked questions

Provider availability, context limits and API capabilities.

Which providers serve Microsoft: Phi 4?

DeepInfra Bf16 are listed in the public reference data. Check CLSSAI availability above before making a request.

What context length is available for Microsoft: Phi 4?

The largest available context window is 16,384 tokens.

Which API parameters does Microsoft: Phi 4 support?

The providers list 13 distinct supported parameters. Open an endpoint row to inspect them.

Explore

Continue comparing models or start integrating.