phi 4

Name: phi 4
Author: Microsoft

by MicrosoftPhi 4

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released December 202416K context≈ 12,288 words14.7B params

Phi-4 punches well above its weight class for a 14B model, showing particular strength in structured reasoning, math, and code — areas where careful training data curation pays off more than raw scale. It tends to be methodical and precise, favoring step-by-step logic over broad generalization. The trade-off is that its world knowledge and conversational range are narrower than larger models.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Instruction Following

Strong

Reasoning & Logic

phi 4

by MicrosoftPhi 4

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released December 202416K context≈ 12,288 words14.7B params

Phi-4 punches well above its weight class for a 14B model, showing particular strength in structured reasoning, math, and code — areas where careful training data curation pays off more than raw scale. It tends to be methodical and precise, favoring step-by-step logic over broad generalization. The trade-off is that its world knowledge and conversational range are narrower than larger models.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Instruction Following

Strong

Reasoning & Logic

Glossary

Data CurationThe process of carefully selecting, cleaning, and organizing training data to improve model quality; better curated data often leads to better model performance.GeneralizationA model's ability to perform well on new, unseen data that differs from what it was trained on.ReasoningThe model's ability to work through multi-step logical problems and provide justified answers rather than just pattern-matching.Structured ReasoningThe ability to follow logical steps and rules systematically to solve problems, often involving breaking down complex tasks into smaller, ordered components.Training DataThe examples and information used to teach a model how to perform a task, in this case human-written and AI-generated grammatical corrections.Training Data CurationThe process of carefully selecting, filtering, and organizing training data to improve a model's performance on specific tasks rather than relying solely on larger datasets.World KnowledgeA model's learned understanding of facts, concepts, and relationships about the real world, typically acquired during training on diverse text data.

phi 4

Capabilities

phi 4

Capabilities

Benchmark Scores

Use Case Fit

Glossary