Microsoft model product
Phi 4
phi-4 is a state-of-the-art open model built to excel at advanced reasoning, coding, and knowledge tasks.
Updated Aug 17, 2026. Default version: Phi 4
LLMBoard Score8.6Phi 4
Coverage100%13 benchmark families
Context window16KTokens
Official input price$0.125Azure API
Phi 4 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Phi 4 Benchmark Results
Benchmark scores for Phi 4.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkPhiBench | Score56.2% | Rank03 | Participants3 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkHumanEval+ | Score82.8% | Rank05 | Participants10 | Percentile55.6% | EvidenceC | Evaluated |
| BenchmarkArena Hard | Score75.4% | Rank09 | Participants26 | Percentile68.0% | EvidenceC | Evaluated |
| BenchmarkDROP | Score75.5% | Rank19 | Participants30 | Percentile37.9% | EvidenceC | Evaluated |
| BenchmarkMATH | Score80.4% | Rank19 | Participants71 | Percentile74.3% | EvidenceC | Evaluated |
| BenchmarkMGSM | Score80.6% | Rank19 | Participants31 | Percentile40.0% | EvidenceC | Evaluated |
| BenchmarkLiveBench | Score47.6% | Rank36 | Participants38 | Percentile5.4% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score82.6% | Rank41 | Participants66 | Percentile38.5% | EvidenceC | Evaluated |
| BenchmarkMMLU | Score84.8% | Rank45 | Participants101 | Percentile56.0% | EvidenceC | Evaluated |
| BenchmarkSimpleQA | Score3.0% | Rank45 | Participants47 | Percentile4.3% | EvidenceC | Evaluated |
| BenchmarkIFEval | Score63.0% | Rank65 | Participants67 | Percentile3.0% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score70.4% | Rank86 | Participants134 | Percentile36.1% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score56.1% | Rank170 | Participants239 | Percentile29.0% | EvidenceC | Evaluated |
Phi 4 Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenatext | Categoryoverall | Rank287 | Rating / score1216.8 | Votes24,126 | ObservationsN/A | Result date |
| Arenatext style control | Categoryoverall | Rank293 | Rating / score1256.2 | Votes24,126 | ObservationsN/A | Result date |
Phi 4 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $0.125 input, $0.50 output per 1M
- Official provider
- Azure
- Lowest third-party
- From $0.07 input, $0.14 output per 1M via OpenRouter
- Tracked offerings
- 3
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderOpenRouter | Provider model IDmicrosoft/phi-4 | Regionglobal | Input / 1M$0.07 | Output / 1M$0.14 | Context16.4K | Updated |
| ProviderAzure Cognitive Services | Provider model IDphi-4 | Regionglobal | Input / 1M$0.125 | Output / 1M$0.50 | Context128K | Updated |
| ProviderAzure | Provider model IDphi-4 | Regionglobal | Input / 1M$0.125 | Output / 1M$0.50 | Context128K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Phi 4 Runtime Performance
Provider-specific output speed and catalog latency for Phi 4. Runtime does not affect the capability score.
Sort by
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderDeepInfra | Output Speed33 tok/s | Catalog Latency0.2 s | Max Input16K | Max Output16K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Phi 4 Specifications
Technical details for the model's default version.
- Version
- Phi 4
- Released
- Dec 12, 2024
- Knowledge cutoff
- Jun 1, 2024
- Parameters
- 14.7B
- Context window
- 16K
- Max output
- 16K
- Inputs
- text
- Outputs
- text
- Open weights
- No
- License
- MIT
Phi 4 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionPhi 4 | Released | LLMBoard8.6 | Parameters14.7B | Context16K | Max output16K | Open weightsNo | LicenseMIT |
Phi 4 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Phi 4
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
MI
Phi 4 multimodal
Microsoft
MI
Phi 3.5 MoE
Microsoft
MI
Phi 4 Mini Reasoning
Microsoft
MI
Phi 3.5 vision
Microsoft
MI
Phi 3.5 mini
Microsoft
MI
Phi 4 Mini
Microsoft
What is Phi 4?
Key information about Phi 4 and its available data.
phi-4 is a state-of-the-art open model built to excel at advanced reasoning, coding, and knowledge tasks. It leverages a blend of synthetic data, filtered web data, academic texts, and supervised fine-tuning for precision, alignment, and safety.
Data as of 2026-08-17.
FAQ
Common questions about Phi 4.
When was Phi 4 released?
Phi 4's default version was released on Dec 12, 2024.
How much does Phi 4 cost?
Phi 4's official API price is $0.125 per million input tokens and $0.50 per million output tokens via Azure. The lowest tracked third-party offer starts at $0.07 input and $0.14 output via OpenRouter.
Who created Phi 4?
Phi 4 was created by Microsoft.
What is the context window for Phi 4?
The default version has a 16K token context window.
Is Phi 4 open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer Phi 4?
3 provider offerings are linked to the default version.
What models should I compare Phi 4 with?
Nearby ranked alternatives include Llama 3.2 90B, Llama 3.1 Nemotron Nano 8B, Qwen2 72B.